Every source behind the read

111 sources in All practitioners, with product documentation Showing all 111 sources

1. Negative

   port11, Hacker News commenter

   Comment on: JetBrains reports revenue growth, net financial loss for 2025Hacker News comment

   Calls Junie mediocre at best and says the JetBrains AI plan quota runs out in three days.

   Published Oct 8, 2026self-described: WebStorm user

   Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

   Origin

   Published practitioner account

   Checked

   Pseudonymous; Customer of Junie; Used it directly; Specific details

   Role

   Pseudonymous account; identity not pursued

   Relationship

   customer (self disclosed)

   How they know

   Used it directly

   Support

   No artifact or independent corroboration recorded

   Date and version

   2026-10-08

   Not counted

   Cost predictability and limits, limits in practice: repeats a stronger account in its group

   Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

   Counts toward

   • Overall signal
   • Code Quality: Mixed or negative
   • Credits & Cost: Raised as a concern

   Published on

   Developer Communities

   Who

   Individual practitioner

   Evidence

   Reported

   Applies to

   Current

   Observed on

   Junie with JetBrains AI plan

2. Mixed

   dools, Hacker News commenter

   Comment on: JetBrains reports revenue growth, net financial loss for 2025Hacker News comment

   Junie in IntelliJ was what convinced them AI coding was the future. They later ran Junie CLI in parallel with other agents but dropped it because instruction compliance broke down on their workflows.

   Published Oct 6, 2026self-described: Kotlin Multiplatform projects; a team sharing JetBrains AI credits

   Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

   Origin

   Published practitioner account

   Checked

   Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Observed on OpenAI GPT 5

   Role

   Pseudonymous account; identity not pursued

   Relationship

   No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

   How they know

   Used it directly

   Support

   No artifact or independent corroboration recorded

   Date and version

   2026-10-06, OpenAI GPT 5

   Not counted

   Delivered outcomes, task completion: repeats a stronger account in its group

   Not counted

   Operational control, models and deployment: repeats a stronger account in its group

   Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

   Counts toward

   • Overall signal
   • Test Writing: Positive
   • Inconsistent Results: Raised as a concern
   • Needs Close Steering: Raised as a concern
   • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

   Published on

   Developer Communities

   Who

   Very small team

   Evidence

   Reported

   Applies to

   Current

   Observed on

   Junie in IntelliJ IDEA, later Junie CLI

3. Mixed

   cmrdporcupine, Hacker News commenter

   Comment on: JetBrains reports revenue growth, net financial loss for 2025Hacker News comment

   Calls Junie really good, but JetBrains restricts it to a curated list of model variants without support for arbitrary OpenAI-compatible endpoints. Its AI strategy has pivoted too often.

   Published Oct 6, 2026self-described: JetBrains user for 25+ years

   Practitioner account, pseudonymous; direct use; counts toward Operational control How GILD checked it

   Origin

   Published practitioner account

   Checked

   Pseudonymous; Customer of Junie; Used it directly; Specific details

   Role

   Pseudonymous account; identity not pursued

   Relationship

   customer (self disclosed)

   How they know

   Used it directly

   Support

   No artifact or independent corroboration recorded

   Date and version

   2026-10-06

   Counts toward

   Operational control, models and deployment: unfavorable, G2 (2 units before bounds)

   Checked Oct 30, 2026 under gild-policy/1.0.0. The scorecard

   Counts toward

   • Overall signal
   • Code Quality: Positive
   • Credits & Cost: Discussed without concern

   Published on

   Developer Communities

   Who

   Individual practitioner

   Evidence

   Reported

   Applies to

   Current

4. Positive

   lowercased, Hacker News commenter

   Comment on: Best LLM for every budget, updated dailyHacker News comment

   Pays US$30 a month for US$35 in credits and, using Junie as a daily driver, tops up about US$20 more on average. The CLI made it more usable.

   Published Sep 24, 2026self-described: Pays own API costs, Junie $30/month plan for local dev work

   Practitioner account, pseudonymous; direct use; counts toward Cost predictability and limits How GILD checked it

   Origin

   Published practitioner account

   Checked

   Pseudonymous; Customer of Junie; Used it directly; Specific details

   Role

   Pseudonymous account; identity not pursued

   Relationship

   customer (self disclosed)

   How they know

   Used it directly

   Support

   No artifact or independent corroboration recorded

   Date and version

   2026-09-24

   Counts toward

   Cost predictability and limits, cost transparency: favorable, G2 (2 units before bounds)

   Counts toward

   Cost predictability and limits, cost predictability: mixed, G2 (2 units before bounds)

   Counts toward

   Cost predictability and limits, limits in practice: mixed, G2 (2 units before bounds)

   Checked Oct 30, 2026 under gild-policy/1.0.0. The scorecard

   Counts toward

   • Overall signal
   • Credits & Cost: Discussed without concern

   Published on

   Developer Communities

   Who

   Individual practitioner

   Evidence

   Measured

   Applies to

   Current

   Observed on

   Junie, US$30 plan with top-ups, CLI

5. Negative

   delbronski, Hacker News commenter

   Comment on: JetBrains Air: A System of Products for Agentic Software DevelopmentHacker News comment

   Gave Junie a solid try but found it always too far behind competing agents, and decided not to renew a JetBrains subscription held for over ten years.

   Published Sep 22, 2026self-described: Paying JetBrains customer for 10+ years, did not renew this year

   Practitioner account, pseudonymous; direct use; not counted, takes no position How GILD checked it

   Origin

   Published practitioner account

   Checked

   Pseudonymous; Customer of Junie; Used it directly

   Role

   Pseudonymous account; identity not pursued

   Relationship

   customer (self disclosed)

   How they know

   Used it directly

   Support

   No artifact or independent corroboration recorded

   Date and version

   2026-09-22

   Not counted

   no aspect: excluded, takes no position on any assessed aspect

   Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

   Counts toward

   • Overall signal
   • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

   Published on

   Developer Communities

   Who

   Individual practitioner

   Evidence

   Reported

   Applies to

   Current

6. Mixed

   GordonS, Hacker News commenter

   Comment on: JetBrains Air: A System of Products for Agentic Software DevelopmentHacker News comment

   Used Junie for small refactorings, where it did a good job quickly, but it handled more complex requests poorly and its UI was tricky. Now uses Codex instead.

   Published Sep 22, 2026self-described: Uses JetBrains Rider and DataGrip regularly

   Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

   Origin

   Published practitioner account

   Checked

   Pseudonymous; No tie to Junie found in GILD's check; Used it directly

   Role

   Pseudonymous account; identity not pursued

   Relationship

   No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

   How they know

   Used it directly

   Support

   No artifact or independent corroboration recorded

   Date and version

   2026-09-22

   Not counted

   Delivered outcomes, task completion: repeats a stronger account in its group

   Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

   Counts toward

   • Overall signal
   • IDE Integration: Mixed or negative
   • Inconsistent Results: Raised as a concern
   • Strongest on scoped work
   • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

   Published on

   Developer Communities

   Who

   Individual practitioner

   Evidence

   Reported

   Applies to

   Current

7. Negative

   backwardsponcho, Hacker News commenter

   Comment on: JetBrains Air: A System of Products for Agentic Software DevelopmentHacker News comment

   Uses Junie at work and is not impressed. It will not run inside an Alpine Linux VM where every other agent they tried worked.

   Published Sep 22, 2026self-described: Junie at work

   Practitioner account, pseudonymous; direct use; counts toward Adoption and upkeep How GILD checked it

   Origin

   Published practitioner account

   Checked

   Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Specific details

   Role

   Pseudonymous account; identity not pursued

   Relationship

   No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

   How they know

   Used it directly

   Support

   No artifact or independent corroboration recorded

   Date and version

   2026-09-22

   Counts toward

   Adoption and upkeep, setup: unfavorable, G2 (2 units before bounds)

   Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

   Counts toward

   • Overall signal
   • Stability: Raised as a concern
   • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

   Published on

   Developer Communities

   Who

   Individual practitioner

   Evidence

   Reported

   Applies to

   Current

8. Positive

   ccb_pnpm, Developer comparing three agent setups

   Same model, very different results: is the agent harness (Junie vs Claude Code vs Hermes) doing more work than the model itself?Reddit post

   Running the same tasks through Hermes, Junie and Claude Code, Junie consistently gave the most satisfying results: fast, finds the real problem, and avoids unrequested changes.

   Published Sep 3, 2026self-described: Same coding tasks run through Hermes, Junie in IntelliJ and Claude Code

   Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes How GILD checked it

   Origin

   Published practitioner account

   Checked

   Pseudonymous; No tie to Junie found in GILD's check; Used it directly

   Role

   Pseudonymous account; identity not pursued

   Relationship

   No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

   How they know

   Used it directly

   Support

   No artifact or independent corroboration recorded

   Date and version

   2026-09-03

   Counts toward

   Delivered outcomes, task completion: favorable, G1 (1 unit before bounds)

   Counts toward

   Delivered outcomes, correctness: favorable, G1 (1 unit before bounds), capped: C-relevance

   Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

   Counts toward

   • Overall signal
   • Codebase Understanding: Positive
   • Code Quality: Positive
   • IDE Integration: Positive
   • Among practitioners who use Junie and another agent, which works better as the main tool?: Junie as the main tool

   Published on

   Developer Communities

   Who

   Individual practitioner

   Evidence

   Reported

   Applies to

   Current

   Observed on

   Junie (Claude-based) in IntelliJ

9. No opinion stated

   wcampospro, GitHub user

   Junie CLI: "The task was stopped due to an error" for all OpenRouter BYOK modelsGitHub issue on JetBrains/junie

   Every OpenRouter model tried through Junie CLI's own-key option failed at once with a generic error, while JetBrains AI models worked in the same install.

   "Native JetBrains AI-provided models work fine; the issue is isolated to the OpenRouter BYOK path."

   Published Aug 4, 2026self-described: Junie CLI on Linux, run from JetBrains Rider's terminal

   Practitioner account, pseudonymous; structured test; counts toward Reliability, Operational control How GILD checked it

   Origin

   Published practitioner account

   Checked

   Pseudonymous; No tie to Junie found in GILD's check; Ran a structured test; Specific details; Observed on 26.7.27 (2470.4)

   Role

   Pseudonymous account; identity not pursued

   Relationship

   No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

   How they know

   Ran a structured test

   Support

   No artifact or independent corroboration recorded

   Date and version

   2026-08-04, 26.7.27 (2470.4)

   Counts toward

   Reliability, run stability: unfavorable, G1 (1 unit before bounds), capped: C-relevance

   Counts toward

   Operational control, models and deployment: unfavorable, G2 (2 units before bounds)

   Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

   Counts toward

   • Stability: Raised as a concern

   Published on

   Developer Communities

   Who

   Individual practitioner

   Evidence

   Reported

   Applies to

   Current

   Observed on

   Junie CLI 26.7.27 with an OpenRouter key; Rider 2026.2

10. Positive

    FpUser, Hacker News commenter

    Comment on: Google Books (or similar) all book scans: $200k bounty (2025)Hacker News comment

    Uses Junie professionally with Gemini Flash, which handles what they ask. More advanced models from other vendors used up credits with no extra benefit.

    Published Jul 4, 2026self-described: Runs their own business

    Practitioner account, pseudonymous; direct use; counts toward Operational control How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Specific details; Observed on Gemini Flash

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2026-07-04, Gemini Flash

    Counts toward

    Operational control, models and deployment: favorable, G2 (2 units before bounds)

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Credits & Cost: Discussed without concern

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Current

    Observed on

    Junie with Gemini Flash

11. No opinion stated

    molbal, GitHub user

    Bug Report: events.jsonl gets too big and fills up disk spaceGitHub issue on JetBrains/junie

    Junie wrote every line a long-running script printed into its session log with no size limit, and in one run the log kept growing until the computer ran out of disk space.

    "The app keeps writing every line of text to the file until the computer completely runs out of disk space."

    Published Jul 4, 2026self-described: Long-running scripts executed by Junie

    Practitioner account, pseudonymous; direct use; counts toward Reliability How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Specific details

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    1 artifact (screenshot)

    Date and version

    2026-07-04

    Counts toward

    Reliability, run stability: unfavorable, G2 (2 units before bounds)

    Checked Oct 30, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Stability: Raised as a concern

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Needs a recheck

    Observed on

    Junie, July 2026

12. Negative

    gquittet, Long-time JetBrains user

    JetBrains AI (Chat & Junie): My feedback after 2 monthsReddit post

    After two months on AI Pro, credits drained fast and Junie became unreliable, implementing a trivial benchmark after an agreed plan, so they downgraded and moved to OpenCode.

    Published Jun 11, 2026self-described: Two months on JetBrains AI Pro

    Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes, Reliability, Cost predictability and limits How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Specific details; Observed on Gemini Flash High

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2026-06-11, Gemini Flash High

    Counts toward

    Cost predictability and limits, cost predictability: unfavorable, G2 (2 units before bounds)

    Counts toward

    Reliability, service availability: unfavorable, G1 (1 unit before bounds)

    Counts toward

    Delivered outcomes, task completion: unfavorable, G2 (2 units before bounds)

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Planning & Control: Mixed or negative
    • Credits & Cost: Raised as a concern
    • Inconsistent Results: Raised as a concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Current

    Observed on

    Junie with Gemini Flash High, AI Pro

13. No opinion stated

    dustin-auby, GitHub user

    Junie ignores ask mode and applies chnagesGitHub issue on JetBrains/junie

    Reports that Junie increasingly edits code while in ask mode and does not show the changes it applied, so he cannot roll them back from Junie.

    "…when using ask mode Junie will take your prompt and apply changes to the code base, further to that it will not show you the applied changes so you cannot rollback…"

    Published May 7, 2026self-described: Uses Junie's ask mode on an existing codebase

    Practitioner account, pseudonymous; direct use; counts toward Operational control How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2026-05-07

    Counts toward

    Operational control, execution control: unfavorable, G1 (1 unit before bounds)

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Planning & Control: Mixed or negative
    • Stability: Raised as a concern

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Needs a recheck

    Observed on

    Junie ask mode, May 2026

14. Negative

    RevEng, Hacker News commenter, self-described

    Comment on: SpaceX says it has agreement to acquire Cursor for $60BHacker News comment

    Used Junie when it first came out and found it okay but unreliable. Now uses Cursor, though they miss PyCharm.

    Published Apr 25, 2026

    Practitioner account, pseudonymous; direct use; not counted, takes no position How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2026-04-25

    Not counted

    no aspect: excluded, takes no position on any assessed aspect

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Overall signal
    • Inconsistent Results: Raised as a concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Needs a recheck

15. Positive

    FpUser, Hacker News commenter

    Comment on: SpaceX says it has agreement to acquire Cursor for US$60BHacker News comment

    Uses Junie with the bundled credits plus top-ups costing about US$100 a month, billed to clients. Says the experience is excellent and far better than Cursor.

    Published Apr 22, 2026self-described: Client work in CLion, WebStorm and PyCharm

    Practitioner account, pseudonymous; direct use; not counted, takes no position How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; Customer of Junie; Used it directly; Observed on all you can eat subscription

    Role

    Pseudonymous account; identity not pursued

    Relationship

    customer (self disclosed)

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2026-04-22, all you can eat subscription

    Not counted

    no aspect: excluded, takes no position on any assessed aspect

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Credits & Cost: Discussed without concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Junie as the main tool

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Current

    Observed on

    Junie in CLion, WebStorm and PyCharm; All Products subscription plus credit top-ups

16. Positive

    dzhiurgis, Hacker News commenter

    Comment on: SpaceX says it has agreement to acquire Cursor for US$60BHacker News comment

    Uses Gemini through Junie and finds it far cheaper than Claude, faster and better at following instructions, with the IDE symbol table making everything just work.

    Published Apr 22, 2026self-described: Uses Junie with Gemini in a JetBrains IDE

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Observed on Gemini

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2026-04-22, Gemini

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Not counted

    Delivered outcomes, context handling: repeats a stronger account in its group

    Checked Oct 30, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Code Quality: Positive
    • IDE Integration: Positive
    • Value for Money: Positive
    • Credits & Cost: Discussed without concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Junie as the main tool

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Current

    Observed on

    Junie with Gemini

17. Mixed

    delbronski, Hacker News commenter, self-described

    Comment on: Our eighth generation TPUs: two chips for the agentic eraHacker News comment

    Junie works best when the engineer stays involved and names the files and the change. It is faster and to the point there, but breaks when asked to code a whole thing.

    Published Apr 22, 2026

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2026-04-22

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Inconsistent Results: Raised as a concern
    • Strongest on scoped work
    • How much should Junie complete before a human checkpoint?: Tighter steering

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Current

    Observed on

    Junie with Gemini

18. Mixed

    KronisLV, Hacker News commenter

    Comment on: Ask HN: How is AI-assisted coding going for you professionally?Hacker News comment

    Found JetBrains AI and Junie, bundled with the IDEs they pay for, nice but limited by quotas, and has since moved to Claude Code on a Max plan.

    Published Mar 16, 2026self-described: Software engineer in Latvia, pays for own JetBrains IDEs

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; Customer of Junie; Used it directly

    Role

    Pseudonymous account; identity not pursued

    Relationship

    customer (self disclosed); customer (self disclosed)

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2026-03-16

    Not counted

    Cost predictability and limits, limits in practice: repeats a stronger account in its group

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Credits & Cost: Raised as a concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie bundled with the All Products Pack

19. Negative

    Ken Kousen, Independent trainer, Kousen IT

    Tales from the jar side: Puppies and Mockito, Anthropic settlement, 1 MILLION tokens, Junie Air and CLI, FB and Moltbook: a Match made in Hades, and social media stuffNewsletter, Tales from the jar side (Kousen IT newsletter, Substack)

    Finds IDE Junie great at planning but too slow. The CLI is faster but ran ahead of planning mode, and sees no reason to pick it over Codex, Gemini or Claude Code.

    Published Mar 15, 2026self-described: One of the author's existing projects

    Practitioner account, self-described; direct use; counts toward Delivered outcomes, Reliability, Operational control How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; Used it directly; Observed on Gemini 3 Flash

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2026-03-15, Gemini 3 Flash

    Counts toward

    Delivered outcomes, task completion: unfavorable, G1 (1 unit before bounds), capped: C-relationship

    Counts toward

    Reliability, run stability: unfavorable, G1 (1 unit before bounds), capped: C-relationship

    Counts toward

    Operational control, execution control: unfavorable, G1 (1 unit before bounds), capped: C-relationship

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Planning & Control: Mixed or negative
    • Speed: Raised as a concern
    • Needs Close Steering: Raised as a concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

    Published on

    Engineering Blogs & Reviews

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Current

    Observed on

    Junie CLI with default Gemini 3 Flash; Junie in IntelliJ IDEA, PyCharm and WebStorm

20. Positive

    Artur Skowronski, Head of Java / Kotlin Engineering at VirtusLab

    JetBrains launches Air and Junie CLI - JVM Weekly vol. 166Newsletter, JVM Weekly (Substack)

    After Claude Code's TUI garbled over Mosh on a phone, switched to Junie CLI, whose simpler interface worked for fire-and-forget agent sessions checked remotely.

    Published Mar 12, 2026self-described: Personal setup: Junie CLI over Tailscale, Mosh and tmux from an iPhone

    Practitioner account, self-described; direct use; counts toward Reliability, Operational control How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; No tie to Junie found in GILD's check; Used it directly; Specific details; Observed on beta

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, author profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2026-03-12, beta

    Counts toward

    Reliability, run stability: favorable, G2 (2 units before bounds)

    Counts toward

    Operational control, models and deployment: mixed, G2 (2 units before bounds)

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Credits & Cost: Discussed without concern
    • Evaluated as an engineering agent
    • How much should Junie complete before a human checkpoint?: Broader delegation

    Published on

    Engineering Blogs & Reviews

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Current

    Observed on

    Junie CLI (beta)

21. Positive

    jacobn, Hacker News poster; runs commercial and pro-bono utility sites with a team

    Show HN: Tensor Spy: inspect NumPy and PyTorch tensors in the browser, no uploadHacker News Show HN post

    Their team built tensorspy.com with Junie and ChatGPT 5.2 in about two weeks, a project they say would have taken months without agentic coding.

    Published Mar 2, 2026self-described: Small team side project (tensorspy.com)

    Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Specific details

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2026-03-02

    Counts toward

    Delivered outcomes, task completion: favorable, G2 (2 units before bounds)

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Faster on self-directed work
    • Evaluated as an engineering agent

    Published on

    Developer Communities

    Who

    Very small team

    Evidence

    Reported

    Applies to

    Current

    Observed on

    Junie with ChatGPT 5.2

22. Mixed

    soco, Hacker News commenter, self-described

    Comment on: Anthropic Claude Max US$200/mo: They claim 99% uptime, I calculated 84% Loss: US$780Hacker News comment

    Gets value from Junie but reports frequent agent crashes that waste tokens, occasional short outages, and spending of US$10 or more on a normal coding day even with Gemini 3 Flash.

    Published Feb 4, 2026

    Practitioner account, pseudonymous; direct use; counts toward Reliability How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Observed on Gemini 3 Flash

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2026-02-04, Gemini 3 Flash

    Counts toward

    Reliability, run stability: unfavorable, G1 (1 unit before bounds)

    Counts toward

    Reliability, service availability: unfavorable, G1 (1 unit before bounds)

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Value for Money: Positive
    • Credits & Cost: Raised as a concern
    • Stability: Raised as a concern

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Current

    Observed on

    Junie with Gemini 3 Flash (default) and other LLMs

23. Positive

    Victor Bash, Independent developer (hackathon entrant)

    I built a browser game for the Cloud9 x JetBrains Hackathon and it started from a really simple…Personal blog post, Medium

    Leaned on Junie throughout a hackathon game build for boilerplate, refactoring and bug reasoning, such as diagnosing arrow tunneling, and felt the IDE plus agent kept the work in flow.

    Published Jan 25, 2026self-described: Hackathon browser game in WebStorm (TypeScript, Babylon.js)

    Practitioner account, self-described; opinion; not counted, takes no position How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; Opinion

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Opinion, no use stated

    Support

    No artifact or independent corroboration recorded

    Date and version

    2026-01-25

    Not counted

    no aspect: excluded, takes no position on any assessed aspect

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Codebase Understanding: Positive
    • IDE Integration: Positive
    • Faster on self-directed work

    Published on

    Engineering Blogs & Reviews

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Current

    Observed on

    Junie in WebStorm

24. Mixed

    atraac, Hacker News commenter

    Comment on: Ask HN: Do you have any evidence that agentic coding works?Hacker News comment

    Uses Junie only for easy, neatly defined tasks such as readmes, scripts and one-off tools. Would not trust it with day-to-day features and must verify its output thoroughly.

    Published Jan 21, 2026self-described: Backend work; one-off tools and small utilities

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2026-01-21

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Inconsistent Results: Discussed without concern
    • Requires Strong Review: Raised as a concern
    • Strongest on scoped work
    • How much should Junie complete before a human checkpoint?: Tighter steering

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Current

    Observed on

    Junie with various underlying models

25. Positive

    Ash Allen, Freelance Laravel web developer

    Using AI to Build a Laravel Feature with Junie in PhpStormPersonal blog post, Ash Allen Design (ashallendesign.co.uk)

    Uses Junie in PhpStorm to debug and write features. A pretty positive experience overall, though it sometimes misunderstands requests and struggled late in the recorded session.

    Published Jan 16, 2026self-described: Laravel application with the Vue.js starter kit in PhpStorm

    Practitioner account; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Used it directly; Specific details

    Role

    Name unknown; employer unknown; role unknown

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Used it directly

    Support

    1 artifact (recording)

    Date and version

    2026-01-16

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Inconsistent Results: Raised as a concern
    • Faster on self-directed work

    Published on

    Engineering Blogs & Reviews

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Current

    Observed on

    Junie in PhpStorm

26. Mixed

    Ash Allen, Freelance Laravel developer and author

    Building a Laravel Feature with Junie in PhpStormVideo, YouTube

    Recorded feature build where Junie does much of the heavy lifting but cannot finish the page after several rounds. The author calls it hit and miss and still needing handholding.

    "…it a prompt and it'll build a feature for you. Maybe it will be able to in the future, but um right now it does need a bit more handholding, but still extremely powerful and I'm looking forward to seeing how it's going in the…"

    Published Jan 14, 2026self-described: Adding a settings page to a fresh Laravel application in PhpStorm

    Practitioner account, self-described; direct use; counts toward Delivered outcomes, Reliability How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; No tie to Junie found in GILD's check; Used it directly; Observed on GPT5 model; Specific details

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, author profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2026-01-14, GPT5 model

    Counts toward

    Delivered outcomes, correctness: unfavorable, G2 (2 units before bounds)

    Counts toward

    Reliability, run stability: unfavorable, G1 (1 unit before bounds)

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Inconsistent Results: Raised as a concern
    • Requires Strong Review: Raised as a concern

    Published on

    Videos & Talks

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Current

    Observed on

    Junie in PhpStorm

27. Mixed

    bpavuk, Hacker News commenter

    Comment on: MCP is a fadHacker News comment

    Delegated ready Figma designs to Junie on Gemini 3 Flash for Jetpack Compose. It did best of three agents tried and used Context7 for newer APIs, but fumbled details, performance, and abstraction.

    Published Jan 10, 2026self-described: Developer in Ukraine, personal projects in GTK-rs, Jetpack Compose and Svelte

    Practitioner account, pseudonymous; direct use; counts toward Operational control How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Specific details; Observed on Gemini 3 Flash

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2026-01-10, Gemini 3 Flash

    Counts toward

    Operational control, models and deployment: favorable, G2 (2 units before bounds)

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Code Quality: Mixed or negative
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Junie as the main tool

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie with Gemini 3 Flash, December 2025

28. Positive

    David Haley, Software and cloud architect, Google Cloud Expert

    Auto-loading .nvmrc in JetBrains Junie terminalDeveloper blog post (dev.to), DEV Community

    Reports an overall great experience with Junie since April 2025, and shares a shell fix after Junie's terminal ignored .nvmrc, which broke test runs in a React project.

    Published Dec 23, 2025self-described: React/TypeScript project

    Practitioner account, self-described; direct use; counts toward Adoption and upkeep How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; Used it directly; Specific details

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-12-23

    Counts toward

    Adoption and upkeep, setup: unfavorable, G2 (2 units before bounds), capped: C-relationship

    Checked Oct 30, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • IDE Integration: Mixed or negative
    • Evaluated as an engineering agent

    Published on

    Engineering Blogs & Reviews

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie in a JetBrains IDE terminal (JediTerm)

29. Mixed

    PaulHoule, Hacker News commenter

    Comment on: AI helps ship faster but it produces 1.7× more bugsHacker News comment

    Describes a mixed experience: Junie shines at explaining codebases and helped them write more tests and better comments, but sometimes produces convoluted code and can slip bad changes past an inattentive user.

    Published Dec 18, 2025self-described: React and TypeScript front end plus backend work

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-12-18

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Not counted

    Delivered outcomes, context handling: repeats a stronger account in its group

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Codebase Understanding: Positive
    • Test Writing: Positive
    • Inconsistent Results: Raised as a concern
    • Requires Strong Review: Raised as a concern
    • Written guidelines make results better

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

30. No opinion stated

    Sebastian Müller, Managing Director at yWorks

    X post: "what's wrong with Junie via IDE Services"X post

    Says Junie via IDE Services, which the company switched to a few days earlier, stopped working for everyone in the company that day, and asks JetBrains whether it is a known issue or a temporary glitch.

    "…today, Junie stopped working for everyone in our company…"

    Published Dec 18, 2025yWorks, a Tübingen diagramming software company that moved its developers to Junie through JetBrains IDE Services

    Practitioner account, pseudonymous; direct use; counts toward Reliability How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Specific details

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile, author profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-12-18

    Counts toward

    Reliability, service availability: unfavorable, G2 (2 units before bounds)

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Stability: Raised as a concern

    Published on

    Developer Communities

    Who

    Technical leadership

    Evidence

    Reported

    Applies to

    Needs a recheck

    Observed on

    Junie provisioned through JetBrains IDE Services

31. Positive

    Mike Lundahl, Developer writing on a personal site

    JetBrains Junie: AI Coding Without the ChaosPersonal blog post, mikelundahl.com

    Uses Junie daily mainly as a pair programmer for questions and debugging, letting it write more only on new projects after a written spec and guidelines. Finds it slow at times.

    Published Dec 11, 2025self-described: Mike Lundahl, mikelundahl.com

    Practitioner account, self-described; direct use; counts toward Delivered outcomes, Operational control How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; Used it directly

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-12-11

    Counts toward

    Delivered outcomes, task completion: mixed, G1 (1 unit before bounds), capped: C-relationship

    Counts toward

    Delivered outcomes, correctness: favorable, G1 (1 unit before bounds), capped: C-relationship

    Counts toward

    Delivered outcomes, context handling: favorable, G1 (1 unit before bounds), capped: C-relationship

    Counts toward

    Operational control, execution control: favorable, G1 (1 unit before bounds), capped: C-relationship

    Counts toward

    Operational control, models and deployment: unfavorable, G1 (1 unit before bounds), capped: C-relationship

    Not counted

    Operational control, data control: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Codebase Understanding: Positive
    • Planning & Control: Positive
    • Speed: Raised as a concern
    • Written guidelines make results better
    • Faster on self-directed work
    • How much should Junie complete before a human checkpoint?: Tighter steering

    Published on

    Engineering Blogs & Reviews

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie alongside JetBrains AI Assistant in a JetBrains IDE

32. Positive

    Somasundaram Mahesh, Android developer

    How to use JetBrains Junie in your Terminal (First Look)Video, YouTube

    First look at Junie CLI on an Android codebase: it read the file structure, summarized the UI architecture and proposed a concrete quick-win refactor.

    "Empty or error states we can handle. Navigation and state, accessibility, performance and polish, design and consistency, interaction testing, quick thing. It has also made a quick win like. directly with the code how we can add the empty state and so on. So yeah,…"

    Published Nov 27, 2025self-described: Android MVVM template project, Junie CLI

    Practitioner account, self-described; direct use; counts toward Operational control How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; No tie to Junie found in GILD's check; Used it directly; Specific details

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, author profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-11-27

    Counts toward

    Operational control, models and deployment: favorable, G2 (2 units before bounds)

    Not counted

    Delivered outcomes, context handling: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Codebase Understanding: Positive

    Published on

    Videos & Talks

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie CLI with a JetBrains account; Sonnet, GPT 5.1 and Gemini 3 listed

33. Positive

    Alessandro Giacomella, Developer who sets AI-use rules for their team

    Coding in the Age of AI Agents: what I learned after letting them write (some of) my codePersonal blog post, Medium

    After gating AI code generation for their team, reopened it with guardrails and describes Junie as the careful agent they use for safe refactoring that follows project patterns.

    Published Nov 21, 2025self-described: Team codebase

    Practitioner account; structured test; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Ran a structured test

    Role

    Name unknown; employer unknown; role unknown

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Ran a structured test

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-11-21

    Counts toward

    Delivered outcomes, correctness: favorable, G1 (1 unit before bounds), capped: C-relationship, C-unknown-control

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Code Quality: Positive
    • Planning & Control: Positive
    • Requires Strong Review: Discussed without concern
    • How much should Junie complete before a human checkpoint?: Tighter steering

    Published on

    Engineering Blogs & Reviews

    Who

    Technical leadership

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie in a JetBrains IDE, alongside Cursor, Claude Code and GitHub Copilot

34. Positive

    cmpalmer52, Hacker News commenter

    Comment on: Vibe Code Warning - A personal casestudyHacker News comment

    Built a pull-request dashboard for their team with Junie in WebStorm in about three hours of step-by-step tasks. The team uses it daily, but Junie misses things in their huge main codebase.

    Published Nov 10, 2025self-described: Team tooling: pull-request dashboard across Azure DevOps and YouTrack, built in WebStorm

    Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Specific details

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-11-10

    Counts toward

    Delivered outcomes, task completion: favorable, G2 (2 units before bounds)

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Codebase Understanding: Mixed or negative
    • Written guidelines make results better
    • Faster on self-directed work
    • Strongest on scoped work
    • How much should Junie complete before a human checkpoint?: Tighter steering

    Published on

    Developer Communities

    Who

    Very small team

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie in WebStorm, with ChatGPT 5 for architecture

35. Mixed

    Umayanga Gunawardhana, Developer (Medium author)

    Taming the AI Beast: How Hierarchical Planning with Backtracking Unlocks Junie's Full Potential for Epic Coding TasksPersonal blog post, Medium

    Finds Junie strong on small tweaks but prone to drift on large multi-module tasks, and shares a guidelines.md prompt to keep its plans on track at extra token cost.

    Published Nov 9, 2025self-described: Golang projects

    Practitioner account, self-described; direct use; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; Used it directly

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-11-10

    Counts toward

    Delivered outcomes, context handling: unfavorable, G1 (1 unit before bounds), capped: C-relationship

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Credits & Cost: Discussed without concern
    • Inconsistent Results: Raised as a concern
    • Written guidelines make results better
    • Strongest on scoped work

    Published on

    Engineering Blogs & Reviews

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

36. Mixed

    VRT303, Developer who tracks their team's AI usage and helps junior developers

    Comment on: Can anyone share their experience with Junie VS VS code / cursor?Reddit comment

    Comparing Junie and Cursor on the same model and prompts across their team, Junie was slower but made fewer mistakes and got changes right in one go. The UI is clunky.

    Published Nov 8, 2025self-described: Team on huge mature projects; GoLand user

    Practitioner account, pseudonymous; structured test; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Ran a structured test

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Ran a structured test

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-11-08

    Counts toward

    Delivered outcomes, correctness: favorable, G1 (1 unit before bounds)

    Counts toward

    Delivered outcomes, task completion: mixed, G1 (1 unit before bounds)

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Code Quality: Positive
    • IDE Integration: Mixed or negative
    • Speed: Raised as a concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Junie as the main tool

    Published on

    Developer Communities

    Who

    Very small team

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie and AI Assistant in GoLand, compared with Cursor on the same model

37. Positive

    Miguel Teheran, Software developer and content creator

    Automating Code Reviews with Junie by JetBrainsDeveloper blog post (dev.to), DEV Community

    Used Junie in Rider to review a branch with planted issues. It found and fixed them plus older problems nearby, which they call an exceptional code review tool.

    Published Oct 21, 2025self-described: Practice C# minimal API project in Rider

    Practitioner account, self-described; structured test; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; Ran a structured test; Specific details; Needs recheck after 12 months

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Ran a structured test

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-10-21, needs a recheck

    Counts toward

    Delivered outcomes, task completion: favorable, G1 (1 unit before bounds), capped: C-relevance, C-relationship

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Codebase Understanding: Positive
    • Evaluated as an engineering agent

    Published on

    Engineering Blogs & Reviews

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie in Rider

38. Positive

    Mirzamehdi Karimov, Kotlin Multiplatform developer (hackathon participant)

    How to Get the Most Out of Junie in a Kotlin Multiplatform ProjectPersonal blog post, Medium

    Built two KMP apps with Junie. Code rarely failed to compile but tasks were slow, so they ran two projects in parallel and improved results with Sonnet and evolving guidelines.

    Published Oct 19, 2025self-described: Two Compose/Kotlin Multiplatform apps built during a hackathon with free Junie access

    Practitioner account, self-described; direct use; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; Used it directly; Needs recheck after 12 months

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-10-19, needs a recheck

    Counts toward

    Delivered outcomes, task completion: favorable, G1 (1 unit before bounds), capped: C-relationship

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Code Quality: Positive
    • Speed: Raised as a concern
    • Written guidelines make results better
    • Faster on self-directed work
    • How much should Junie complete before a human checkpoint?: Broader delegation

    Published on

    Engineering Blogs & Reviews

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie, default GPT-5 switched to Claude Sonnet

39. Positive

    Arghya Banerjee, Developer with under two years in IT

    How AI Blurred My Thinking, and How I'm Winning It BackPersonal blog post, Medium

    Compared a manual refactor with a Junie-assisted one under TDD guidelines. Junie was much faster and generated tests, while the developer kept ownership of architecture and reviewed tests.

    Published Oct 18, 2025self-described: Refactoring a monolithic Python script from a live project

    Practitioner account; structured test; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Ran a structured test; Needs recheck after 12 months; Specific details

    Role

    Name unknown; employer unknown; role unknown

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Ran a structured test

    Support

    1 artifact (benchmark output)

    Date and version

    2025-10-18, needs a recheck

    Counts toward

    Delivered outcomes, task completion: favorable, G1 (1 unit before bounds), capped: C-relationship, C-unknown-control

    Counts toward

    Delivered outcomes, correctness: mixed, G1 (1 unit before bounds), capped: C-relationship, C-unknown-control

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Test Writing: Positive
    • Requires Strong Review: Discussed without concern
    • Faster on self-directed work
    • How much should Junie complete before a human checkpoint?: Tighter steering

    Published on

    Engineering Blogs & Reviews

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

40. Mixed

    Steven Webber, Software Engineer at Equal Experts

    Prompts, pitfalls, and progress: What I learned building an app with AI agentsCompany engineering blog post, Equal Experts

    An Equal Experts engineer built a full app through agents without writing code. He started with Junie and found it well integrated with the IDE, but moved to Warp's terminal workflow, and agents often ignored his .junie/guidelines.md rules.

    "…while Junie was great and nicely integrated with the IDE. I ended up preferring the workflow from the terminal as offered by Warp and Amazon Q."

    "Despite setting up a .junie/guidelines.md with explicit instructions (like always updating prompts.txt), agents often ignored them or followed them inconsistently."

    Published Oct 6, 2025Equal Experts AI experiment: a Java Quarkus and React app with CI/CD to AWS, built only through agents and BDD over just over a week

    Practitioner account, name and role on their organization's own page; direct use; counts toward Delivered outcomes, Adoption and upkeep How GILD checked it

    Origin

    Published practitioner account

    Checked

    Name and role on their organization's own page; Used it directly; Needs recheck after 12 months; Specific details

    Role

    Name on a page the person does not control; employer on a page the person does not control; role on a page the person does not control (Software Engineer)

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-10-06, needs a recheck

    Counts toward

    Delivered outcomes, correctness: unfavorable, G2 (2 units before bounds), capped: C-relationship

    Counts toward

    Adoption and upkeep, setup: favorable, G1 (1 unit before bounds), capped: C-relationship

    Checked Oct 30, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • IDE Integration: Positive
    • Needs Close Steering: Raised as a concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

    Published on

    Engineering Blogs & Reviews

    Who

    Professional team

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie with a .junie/guidelines.md file, then Warp and Amazon Q

41. Mixed

    AdComplex1867, Independent developer on JetBrains AI Ultimate

    Junie is now costing me ~US$5 a day (that's about US$1,200/year)Reddit post

    After the credit change exhausted their Ultimate allocation early, they tracked spending daily and measured about US$5 per workday, roughly US$1,200 a year, while still finding value.

    Published Sep 30, 2025self-described: Daily Junie use; tracked credit balance at start and end of each day

    Practitioner account, pseudonymous; direct use; counts toward Cost predictability and limits How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; Customer of Junie; Used it directly; Specific details; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    customer (self disclosed)

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-09-30, ChatGPT-5, needs a recheck

    Counts toward

    Cost predictability and limits, limits in practice: unfavorable, G1 (1 unit before bounds), capped: C-relevance

    Counts toward

    Cost predictability and limits, cost predictability: unfavorable, G1 (1 unit before bounds), capped: C-relevance

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Overall signal
    • Credits & Cost: Raised as a concern

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Measured

    Applies to

    Needs a recheck

    Observed on

    Junie, AI Ultimate plus top-up credits, ChatGPT-5

42. Positive

    Stanislav Radchenko, Android developer at Effective

    How we used Junie by JetBrains to speed up development of an internal projectEngineering blog post, Effective (Medium publication)

    Radchenko reports that Effective used Junie to speed up an internal meeting-room booking app with a Spring Boot backend and Kotlin Multiplatform client, estimating, "in my opinion", about a 30% speed-up and up to 50%. Junie wrote roughly 70/30 of backend code, the reverse on the KMP client, and sometimes looped in complex scenarios.

    Published Sep 19, 2025self-described: Effective, internal Effective Office meeting-room booking app (Kotlin, Spring Boot, Kotlin Multiplatform)

    Practitioner account, self-described; direct use; counts toward Delivered outcomes, Operational control How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; Used it directly; Needs recheck after 12 months; Specific details

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Used it directly

    Support

    1 artifact (transcript)

    Date and version

    2025-09-19, needs a recheck

    Counts toward

    Delivered outcomes, task completion: mixed, G2 (2 units before bounds), capped: C-relationship

    Counts toward

    Delivered outcomes, correctness: unfavorable, G2 (2 units before bounds), capped: C-relationship

    Counts toward

    Delivered outcomes, context handling: unfavorable, G2 (2 units before bounds), capped: C-relationship

    Counts toward

    Operational control, execution control: favorable, G1 (1 unit before bounds), capped: C-relationship

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Overall signal
    • Codebase Understanding: Positive
    • Planning & Control: Positive
    • IDE Integration: Positive
    • Credits & Cost: Discussed without concern
    • Speed: Raised as a concern
    • Inconsistent Results: Raised as a concern
    • Requires Strong Review: Discussed without concern
    • Written guidelines make results better
    • Faster on self-directed work

    Published on

    Engineering Blogs & Reviews

    Who

    Professional team

    Evidence

    Reported

    Applies to

    Needs a recheck

    Observed on

    Junie in IntelliJ IDEA

43. Mixed

    Ken Kousen, Independent trainer who taught an AI Agents with Junie course on O'Reilly

    Tales from the jar side: Junie Joy, Nano Banana, LM Studio, and the usual social media postsNewsletter, Tales from the jar side (Kousen IT newsletter, Substack)

    Teaching a Junie course went better than expected: Junie excels at planning and built small apps quickly, though it is still not their primary tool versus Claude Code and others.

    Published Sep 15, 2025self-described: Live training course exercises in IntelliJ IDEA, PyCharm and WebStorm

    Practitioner account, self-described; direct use; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; Used it directly; Needs recheck after 12 months

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-09-15, needs a recheck

    Counts toward

    Delivered outcomes, task completion: favorable, G1 (1 unit before bounds), capped: C-relevance, C-relationship

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Planning & Control: Positive
    • Speed: Discussed without concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

    Published on

    Engineering Blogs & Reviews

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie in IntelliJ IDEA, PyCharm and WebStorm

44. Negative

    ackfoobar, Hacker News commenter

    Comment on: RubyMine is now free for non-commercial useHacker News comment

    A fan of IntelliJ and Kotlin who used Cursor at work tried Junie on a hobby project and bought a personal Cursor license afterward.

    Published Sep 3, 2025self-described: Cursor at work, Junie on a hobby project

    Practitioner account, pseudonymous; direct use; not counted, takes no position How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-09-03, needs a recheck

    Not counted

    no aspect: excluded, takes no position on any assessed aspect

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

45. Positive

    snuxoll, Hacker News commenter

    Comment on: RubyMine is now free for non-commercial useHacker News comment

    Says Junie benefits from IDE integration: the PSI syntax tree flags invalid code, so after large refactors or feature work the developer can be confident the result is not garbage.

    Published Sep 2, 2025self-described: JetBrains All Products Pack subscriber (AI Pro included)

    Practitioner account, pseudonymous; direct use; counts toward Cost predictability and limits How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Specific details; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-09-02, JetBrains AI Pro, needs a recheck

    Counts toward

    Cost predictability and limits, cost transparency: favorable, G2 (2 units before bounds)

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Checked Oct 30, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Overall signal
    • Code Quality: Positive
    • IDE Integration: Positive
    • Credits & Cost: Discussed without concern
    • How much should Junie complete before a human checkpoint?: Broader delegation

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Needs a recheck

    Observed on

    Junie, JetBrains AI Pro via All Products Pack

46. Mixed

    Rimantas Benetis, Partner, Head of Technology at Gritmind

    We compared 5 AI code assistants - Here's how they stack upCompany blog post, Gritmind Insights

    Gritmind's engineers tested five assistants and found Junie fits teams already on JetBrains IDEs, but its responsiveness was uneven, its autocomplete slow, and it has no checkpoints, so developers must commit first. Cursor and Copilot came out ahead.

    "Initial responsiveness was uneven but has improved with recent updates - still worth monitoring before full adoption…"

    "Lacks checkpoint features - developers must commit before use, increasing the risk of unwanted code changes."

    Published Aug 31, 2025Gritmind, a development partner working inside client stacks; post co-written with Head of Product Laura Graves

    Practitioner account, self-described; direct use; counts toward Delivered outcomes, Operational control, Adoption and upkeep How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; Used it directly; Needs recheck after 12 months

    Role

    Name self-described; employer self-described; role unknown

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-08-31, needs a recheck

    Counts toward

    Delivered outcomes, correctness: unfavorable, G1 (1 unit before bounds), capped: C-relationship

    Counts toward

    Operational control, execution control: unfavorable, G1 (1 unit before bounds), capped: C-relationship

    Counts toward

    Adoption and upkeep, setup: favorable, G1 (1 unit before bounds), capped: C-relationship

    Counts toward

    Delivered outcomes, task completion: mixed, G1 (1 unit before bounds), capped: C-relationship

    Checked Oct 30, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Planning & Control: Mixed or negative
    • IDE Integration: Positive
    • Speed: Raised as a concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

    Published on

    Engineering Blogs & Reviews

    Who

    Technical leadership

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie compared with Cursor, Copilot, AWS Q, and ChatGPT, hands-on as of August 2025

47. Positive

    MarcoAx, Laravel developer (blog author)

    Laravel boost: JetBrains Junie vs Claude Code - Full code review and refactorPersonal blog post (part 4 of a comparison series), Magutti Blog (magutti.com)

    In a blind cross-review, Junie made a simple refactor that worked on the first prompt while Claude Code broke the app. The author found Junie reliable but Claude faster.

    Published Aug 30, 2025self-described: Laravel Boost test project (Google Maps app) built and cross-reviewed by Junie and Claude Code

    Practitioner account; structured test; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Ran a structured test; Specific details; Needs recheck after 12 months

    Role

    Name unknown; employer unknown; role unknown

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Ran a structured test

    Support

    2 artifacts (pull request)

    Date and version

    2025-08-30, needs a recheck

    Counts toward

    Delivered outcomes, task completion: favorable, G1 (1 unit before bounds), capped: C-relationship, C-unknown-control

    Counts toward

    Delivered outcomes, correctness: favorable, G1 (1 unit before bounds), capped: C-relevance, C-relationship, C-unknown-control

    Checked Oct 30, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Codebase Understanding: Positive
    • Code Quality: Positive
    • Speed: Discussed without concern

    Published on

    Engineering Blogs & Reviews

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie in PhpStorm with Laravel Boost, compared with Claude Code

48. Mixed

    smrtinsert, Hacker News commenter, self-described

    Comment on: JetBrains working on higher-abstraction programming languageHacker News comment

    Loves Junie's accuracy, but it works like Claude Code launching a subagent per file, takes as long, likely burns as many tokens, and cannot be switched off.

    Published Aug 14, 2025

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-08-14, needs a recheck

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Checked Oct 30, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Overall signal
    • Code Quality: Positive
    • Speed: Raised as a concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Needs a recheck

49. Negative

    parpfish, Hacker News commenter

    Comment on: Claude is the drug, Cursor is the dealerHacker News comment

    Tried Junie because they love JetBrains IDEs and found it terrible. Cursor was much better at agentic code generation, the change-review workflow and autocomplete.

    Published Aug 11, 2025self-described: JetBrains IDE user; Cursor used on greenfield toy projects

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-08-11, needs a recheck

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Code Quality: Mixed or negative
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

50. Mixed

    Povilas Korop (Laravel Daily), Laravel educator, Laravel Daily

    PhpStorm Junie AI Re-Review: Compared to Cursor / Claude CodeVideo, YouTube

    Re-review after three months: Junie got faster and more stable and is on par with Claude Code and Cursor when it works, but how it tests generated code is a red flag.

    "…cursor if it works. I mean, if specific prompt is executed successfully, but in my testing, it's pretty much okay. What is not okay is some smaller UX issues, some lacking features here and there, but the most danger and kind of red flag is…"

    Published Aug 11, 2025self-described: Laravel practice projects in PhpStorm over a few weeks

    Practitioner account, self-described; direct use; counts toward Operational control, Cost predictability and limits How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; Junie partner: not counted as independent; Used it directly; Needs recheck after 12 months; Specific details

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    sponsor (self disclosed)

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-08-11, Sonet 4, needs a recheck

    Counts toward

    Operational control, execution control: unfavorable, G2 (2 units before bounds)

    Counts toward

    Cost predictability and limits, cost transparency: unfavorable, G2 (2 units before bounds)

    Not counted

    Delivered outcomes, task completion: excluded, praise from someone tied to the vendor, kept as an attributed position

    Not counted

    Reliability, run stability: excluded, praise from someone tied to the vendor, kept as an attributed position

    Not counted

    Operational control, models and deployment: excluded, praise from someone tied to the vendor, kept as an attributed position

    Not counted

    Cost predictability and limits, limits in practice: excluded, praise from someone tied to the vendor, kept as an attributed position

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Overall signal
    • Test Writing: Mixed or negative
    • Credits & Cost: Discussed without concern
    • Speed: Discussed without concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

    Published on

    Videos & Talks

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Needs a recheck

    Observed on

    Junie in PhpStorm, AI Ultimate, Claude Sonnet 4

51. Mixed

    ceronman, Lobsters commenter

    Comment on: GPTs and feeling left behindLobsters comment

    Using GPT-5 in Junie to port a Python test suite to Rust, it honored the no-dependencies request that Gemini CLI ignored and wrote simple code. The result had bugs on complex cases.

    Published Aug 10, 2025self-described: Porting a Python test suite to Rust

    Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-08-10, ChatGPT 5, needs a recheck

    Counts toward

    Delivered outcomes, correctness: mixed, G1 (1 unit before bounds), capped: C-relationship

    Counts toward

    Delivered outcomes, task completion: mixed, G1 (1 unit before bounds), capped: C-relationship

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Code Quality: Positive
    • Inconsistent Results: Raised as a concern

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie with GPT-5

52. Positive

    Adarsh Divakaran, Python developer and author of Deep Dive Python; thanks the JetBrains team for collaborating on the project

    Vibe Coding with Junie - JetBrains AI Coding AgentPersonal blog post, blog.adarshd.dev

    Built an entire Python library with Junie in PyCharm without writing code by hand, praising its planning mode and generated guidelines while advising caution with Brave Mode.

    Published Aug 5, 2025self-described: Side project: Pyzlides, a Python slides-as-code library, in PyCharm

    Practitioner account, self-described; direct use; counts toward Delivered outcomes, Operational control How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; Junie partner: not counted as independent; Used it directly; Specific details; Needs recheck after 12 months

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    partner (suspected); partner (self disclosed)

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-08-05, Claude Sonnet 4, needs a recheck

    Counts toward

    Delivered outcomes, correctness: mixed, G2 (2 units before bounds)

    Counts toward

    Operational control, execution control: mixed, G2 (2 units before bounds)

    Not counted

    Delivered outcomes, task completion: excluded, praise from someone tied to the vendor, kept as an attributed position

    Not counted

    Operational control, models and deployment: excluded, praise from someone tied to the vendor, kept as an attributed position

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Not counted

    Its author says the JetBrains team collaborated on the project the post describes. Kept in the record; it takes part in no figure.

    Published on

    Engineering Blogs & Reviews

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Needs a recheck

    Observed on

    Junie in PyCharm, premium subscription, Claude Sonnet 4, Brave Mode

53. Mixed

    shmatt, Hacker News commenter, self-described

    Comment on: Deep AgentsHacker News comment

    Junie was the first agent they saw with a high-quality to-do list and became their favorite: slow and thoughtful while Cursor rewrote working files. Stopped using it when it became paid.

    Published Aug 1, 2025

    Practitioner account, pseudonymous; direct use; not counted, takes no position How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-08-01, needs a recheck

    Not counted

    no aspect: excluded, takes no position on any assessed aspect

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Overall signal
    • Planning & Control: Positive
    • Credits & Cost: Raised as a concern
    • Speed: Discussed without concern

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Needs a recheck

54. Positive

    manquer, Hacker News commenter, self-described

    Comment on: OpenAI's Windsurf deal is off, and Windsurf's CEO is going to GoogleHacker News comment

    Tried Junie and found it better than Claude Code for some tasks because it uses what the IDE knows, finishing simpler tasks with fewer prompts and tokens than Claude Code or Windsurf.

    Published Jul 13, 2025

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-07-13, needs a recheck

    Not counted

    Delivered outcomes, context handling: repeats a stronger account in its group

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Overall signal
    • IDE Integration: Positive
    • Credits & Cost: Discussed without concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Junie as the main tool

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Needs a recheck

55. Positive

    crappybird, Hacker News commenter, self-described

    Comment on: Apple just released a weirdly interesting coding language modelHacker News comment

    Calls Junie amazing and says only Claude Code and opencode come close, but it burns through credits quickly.

    Published Jul 8, 2025

    Practitioner account, pseudonymous; direct use; not counted, takes no position How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-07-08, needs a recheck

    Not counted

    no aspect: excluded, takes no position on any assessed aspect

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Overall signal
    • Code Quality: Positive
    • Credits & Cost: Raised as a concern

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Needs a recheck

56. Negative

    PaulHoule, Hacker News commenter, self-described

    Comment on: Mercury: Ultra-fast language models based on diffusionHacker News comment

    Found that Junie researched irrelevant things, kept asking permission, and after 15 minutes returned a mountain of broken code, even with actions approved up front.

    Published Jul 7, 2025

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months; Specific details

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-07-07, needs a recheck

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Not counted

    Operational control, execution control: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Overall signal
    • Code Quality: Mixed or negative
    • Speed: Raised as a concern

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Needs a recheck

57. Positive

    Lou Sacco, CTO of DaisyHealth

    Junie AI Code Assistant for WebStorm: The Good, the Bad, and the UglyCTO newsletter post, DaisyHealth's CTO Substack

    After a month of near-daily use in WebStorm, Sacco, DaisyHealth's CTO, calls Junie a legit force multiplier for codebase awareness, unit tests and onboarding, but notes it doesn't always respect framework conventions and that abstract problems lead to painful loops.

    Published Jul 1, 2025self-described: DaisyHealth, JavaScript/TypeScript and NestJS microservices in WebStorm

    Practitioner account, self-described; direct use; counts toward Delivered outcomes, Reliability, Operational control, Cost predictability and limits, Adoption and upkeep How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; Used it directly; Needs recheck after 12 months; Specific details

    Role

    Name self-described; employer self-described; role self-described (DaisyHealth's CTO)

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-07-01, Junie defaults to Sonnet 3.7, but you can switch to Sonnet 4, needs a recheck

    Counts toward

    Delivered outcomes, task completion: favorable, G2 (2 units before bounds), capped: C-relationship

    Counts toward

    Reliability, run stability: unfavorable, G1 (1 unit before bounds), capped: C-relationship

    Counts toward

    Operational control, models and deployment: favorable, G2 (2 units before bounds), capped: C-relationship

    Counts toward

    Cost predictability and limits, cost transparency: favorable, G2 (2 units before bounds), capped: C-relationship

    Counts toward

    Cost predictability and limits, limits in practice: favorable, G2 (2 units before bounds), capped: C-relationship

    Counts toward

    Adoption and upkeep, setup: favorable, G1 (1 unit before bounds), capped: C-relationship

    Not counted

    Delivered outcomes, context handling: repeats a stronger account in its group

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Not counted

    Operational control, execution control: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Overall signal
    • Codebase Understanding: Positive
    • Code Quality: Mixed or negative
    • Planning & Control: Positive
    • Test Writing: Positive
    • Credits & Cost: Discussed without concern
    • Inconsistent Results: Raised as a concern
    • Requires Strong Review: Discussed without concern
    • Faster on self-directed work
    • Strongest on scoped work
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Junie as the main tool

    Published on

    Engineering Blogs & Reviews

    Who

    Technical leadership

    Evidence

    Measured

    Applies to

    Needs a recheck

    Observed on

    Junie in WebStorm, upgraded from AI Pro to AI Ultimate, Claude Sonnet 3.7 default and Sonnet 4

58. Positive

    Amazing_Hospital_515, Senior developer on a team of more than 10 developers

    Junie Ultimate 61 days inReddit post

    After two months on Junie Ultimate in a 10+ developer team, a full-stack feature with e2e tests takes about 2 hours, down from 2-3 days. Colleagues see similar results.

    Published Jun 29, 2025self-described: Monorepo with Spring Boot backend, React/Vite frontend, Playwright e2e tests, infrastructure and Storybook/Figma

    Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes, Cost predictability and limits How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Specific details; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-06-29, Junie Ultimate, needs a recheck

    Counts toward

    Delivered outcomes, task completion: favorable, G2 (2 units before bounds)

    Counts toward

    Delivered outcomes, context handling: favorable, G2 (2 units before bounds)

    Counts toward

    Delivered outcomes, correctness: mixed, G2 (2 units before bounds)

    Counts toward

    Cost predictability and limits, limits in practice: favorable, G2 (2 units before bounds)

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Overall signal
    • Codebase Understanding: Positive
    • Test Writing: Positive
    • Credits & Cost: Discussed without concern
    • Speed: Discussed without concern
    • Faster on self-directed work
    • Strongest on scoped work
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Junie as the main tool

    Published on

    Developer Communities

    Who

    Professional team

    Evidence

    Reported

    Applies to

    Needs a recheck

    Observed on

    Junie Ultimate

59. Negative

    sensanaty, Hacker News commenter

    Comment on: Andrej Karpathy: Software in the era of AI [video]Hacker News comment

    Lists Junie among the agents they and their team tried diligently. For anything beyond toy projects the tools fell apart repeatedly.

    Published Jun 19, 2025self-described: Team trying AI coding tools on non-trivial projects

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-06-19, needs a recheck

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Inconsistent Results: Raised as a concern

    Published on

    Developer Communities

    Who

    Very small team

    Evidence

    Reported

    Applies to

    Still applicable

60. Positive

    kachapopopow, Hacker News commenter

    Comment on: Generative AI coding tools and agents do not work for meHacker News comment

    Had Junie in GoLand build a Kubernetes device plugin for SR-IOV vGPUs while making tea. After reviewing it, the plugin worked nearly first try, saving an hour or more.

    Published Jun 17, 2025self-described: Kubernetes device plugin in Go (GoLand)

    Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Specific details; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-06-17, needs a recheck

    Counts toward

    Delivered outcomes, task completion: favorable, G2 (2 units before bounds)

    Counts toward

    Delivered outcomes, correctness: mixed, G2 (2 units before bounds)

    Checked Oct 30, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Requires Strong Review: Discussed without concern
    • Faster on self-directed work
    • Evaluated as an engineering agent
    • How much should Junie complete before a human checkpoint?: Broader delegation

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie in GoLand

61. Negative

    digitalend, PHP developer at the same company for 15 years

    Your experience with AI Agents and AI Programming Tools in 2025Reddit post

    On a large bespoke PHP codebase with an in-house framework, Junie in PhpStorm (like Claude Code and Aider) did not work at all, misreading routing, ORM and structure.

    Published Jun 17, 2025self-described: Company with an in-house PHP framework and millions of lines of code across separate repos

    Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-06-17, needs a recheck

    Counts toward

    Delivered outcomes, context handling: unfavorable, G1 (1 unit before bounds)

    Counts toward

    Delivered outcomes, task completion: unfavorable, G1 (1 unit before bounds)

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Codebase Understanding: Mixed or negative
    • Inconsistent Results: Raised as a concern

    Published on

    Developer Communities

    Who

    Very small team

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie in PhpStorm

62. Negative

    rolisz, Hacker News commenter

    Comment on: Agentic Coding RecommendationsHacker News comment

    Junie in PyCharm had big problems running tests or even using the project's PyCharm virtual environment, which they expected JetBrains to handle better.

    Published Jun 12, 2025self-described: Python project in PyCharm

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-06-12, needs a recheck

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Test Writing: Mixed or negative
    • IDE Integration: Mixed or negative

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie in PyCharm

63. Mixed

    Michelangelo van Dam, Web application developer and architect

    Personal experience with JetBrains Junie™: day 2Company blog post, in2it

    Pushing Junie to 85 percent PHPUnit coverage, found it ignored guideline test requirements, used deprecated annotations and loosened phpunit.xml strictness to make tests pass.

    Published Jun 11, 2025self-described: PHP training-course registration app (Slim, Twig, PHPUnit) in PhpStorm

    Practitioner account, self-described; direct use; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; Used it directly; Artifact checked (log); Needs recheck after 12 months

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Used it directly

    Support

    1 artifact (log)

    Date and version

    2025-06-11, needs a recheck

    Counts toward

    Delivered outcomes, task completion: mixed, G2 (2 units before bounds), capped: C-relationship

    Counts toward

    Delivered outcomes, correctness: unfavorable, G2 (2 units before bounds), capped: C-relationship

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Code Quality: Mixed or negative
    • Test Writing: Mixed or negative
    • Speed: Discussed without concern
    • Needs Close Steering: Raised as a concern
    • Written guidelines make results better
    • How much should Junie complete before a human checkpoint?: Tighter steering

    Published on

    Engineering Blogs & Reviews

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie in PhpStorm

64. Negative

    aerhardt, Hacker News commenter, self-described

    Comment on: Why agents are bad pair programmersHacker News comment

    Added an instruction to Junie's guidelines document, but Junie still keeps writing a comment every three lines regardless of language.

    Published Jun 10, 2025

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-06-10, needs a recheck

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Checked Oct 30, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Code Quality: Mixed or negative

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

65. Negative

    Vortelf, Go developer

    Junie is disappointing and I'm glad I tried it before buying a blindly subscribing.Reddit post

    On a simple Go struct-tag edit, Junie needed terminal security disabled, edited one line at a time, stopped a third of the way through, and its bug hunt produced breaking suggestions.

    Published Jun 5, 2025self-described: Refactoring a Golang project

    Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes, Operational control How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Specific details; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-06-05, needs a recheck

    Counts toward

    Operational control, execution control: unfavorable, G2 (2 units before bounds)

    Counts toward

    Delivered outcomes, task completion: unfavorable, G2 (2 units before bounds)

    Counts toward

    Delivered outcomes, correctness: unfavorable, G2 (2 units before bounds)

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Codebase Understanding: Mixed or negative
    • Code Quality: Mixed or negative
    • Speed: Raised as a concern
    • Stability: Raised as a concern

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie on a Go project

66. Positive

    okinok, Hacker News commenter

    Comment on: Cursor 1.0Hacker News comment

    Finds Junie comparable to Cursor, and because it comes with the JetBrains license there has been no need to pay for Cursor.

    Published Jun 5, 2025self-described: JetBrains license holder, Junie included in license

    Practitioner account, pseudonymous; direct use; not counted, takes no position How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; Customer of Junie; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    customer (self disclosed)

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-06-05, needs a recheck

    Not counted

    no aspect: excluded, takes no position on any assessed aspect

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Value for Money: Positive
    • Credits & Cost: Discussed without concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Junie as the main tool

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie included with a JetBrains license

67. Mixed

    Calvin Hendryx-Parker, CTO at Six Feet Up

    AI Assistants and uv: Which Dev Tools Deliver?Company news post, Six Feet Up

    Six Feet Up's CTO gave five assistants the same Django refactor. Junie showed promise but felt slower and more rigid than the others, and Claude Code made the most complete refactor.

    "Juni, JetBrains’ AI assistant, showed promise but felt slower and more rigid compared to its peers."

    "Claude Code performed the most complete refactor, modifying dozens of files with minimal prompting and high accuracy."

    Published May 28, 2025Six Feet Up, a Python and AI consultancy; live IndyPy comparison of five assistants converting a Django project's GraphQL API to REST

    Practitioner account, name and role on their organization's own page; second-hand; not counted until substantiated How GILD checked it

    Origin

    Published practitioner account

    Checked

    Name and role on their organization's own page; Second-hand; Specific details; Needs recheck after 12 months

    Role

    Name on a page the person does not control; employer on a page the person does not control; role on a page the person does not control (Six Feet Up CTO)

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Reports others' experience

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-05-28, needs a recheck

    Not counted

    Delivered outcomes, task completion: a lead (G0), counts toward nothing until substantiated

    Checked Oct 30, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Speed: Raised as a concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

    Published on

    Engineering Blogs & Reviews

    Who

    Technical leadership

    Evidence

    Reported

    Applies to

    Needs a recheck

    Observed on

    Junie against Aider, Claude Code, Cursor, and Goose on the same Django refactor prompt

68. Positive

    harrall, Hacker News commenter

    Comment on: Ask HN: Anyone struggling to get value out of coding LLMs?Hacker News comment

    Finds Junie great compared to other offerings: it writes JUnit tests with test doubles, runs them and fixes tests or broken code, though it helps with only about 10% of their work.

    Published May 26, 2025self-described: Java codebase with JUnit tests

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-05-26, needs a recheck

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Test Writing: Positive
    • Evaluated as an engineering agent
    • Strongest on scoped work

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie in IntelliJ

69. Positive

    TrespassersWilliam, Kotlin developer

    Junie is a gamechangerReddit post

    Spent a couple of hours writing .junie/guidelines.md for Compose Multiplatform and Ktor. Junie followed the instructions well, and fixing a gap meant adding one line to the guidelines.

    Published May 25, 2025self-described: Compose Multiplatform frontend and Ktor backend projects

    Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes, Operational control How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-05-25, free tier, needs a recheck

    Counts toward

    Delivered outcomes, correctness: favorable, G1 (1 unit before bounds)

    Counts toward

    Delivered outcomes, task completion: favorable, G1 (1 unit before bounds)

    Counts toward

    Operational control, execution control: favorable, G1 (1 unit before bounds)

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Code Quality: Positive
    • Planning & Control: Positive
    • Written guidelines make results better

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie free tier, .junie/guidelines.md

70. Mixed

    _kidlike, Hacker News commenter

    Comment on: Jules: An asynchronous coding agentHacker News comment

    On a 250k-line Java project, Junie was very good at identifying which parts of the codebase needed to change, but its actual code generation was as poor as any other LLM's.

    Published May 20, 2025self-described: Java project of about 250k lines of code

    Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months; Specific details

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-05-20, needs a recheck

    Counts toward

    Delivered outcomes, context handling: favorable, G2 (2 units before bounds)

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Codebase Understanding: Positive
    • Code Quality: Mixed or negative
    • Planning & Control: Positive

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie on a Java codebase

71. Positive

    Isabel Garrido Cardenas, Senior Software Engineer, Barcelona

    Legacy Code Collaboration with Junie AIPersonal blog post, Isa's code (Hashnode)

    A skeptic turned regular user. In three days Junie helped their team add the first acceptance test to a legacy system that had no tests, but it mishandled unit test doubles and small TDD steps.

    Published May 15, 2025self-described: Their team at work; business-critical legacy Scala/Akka service with no tests

    Practitioner account, self-described; direct use; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; Junie partner: not counted as independent; Used it directly; Specific details; Needs recheck after 12 months

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    program member (self disclosed)

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-05-15, Junie EAP (Early Access Program), needs a recheck

    Counts toward

    Delivered outcomes, task completion: mixed, G1 (1 unit before bounds), capped: C-relationship

    Counts toward

    Delivered outcomes, correctness: mixed, G1 (1 unit before bounds), capped: C-relationship

    Not counted

    Delivered outcomes, context handling: excluded, praise from someone tied to the vendor, kept as an attributed position

    Not counted

    Cost predictability and limits, cost transparency: excluded, praise from someone tied to the vendor, kept as an attributed position

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Codebase Understanding: Positive
    • Planning & Control: Mixed or negative
    • Test Writing: Mixed or negative
    • Inconsistent Results: Raised as a concern
    • Faster on self-directed work

    Published on

    Engineering Blogs & Reviews

    Who

    Very small team

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie EAP

72. Mixed

    Julien Lengrand-Lambert, Google Developer Expert in Kotlin who leads Developer Relations at ING

    My experience using Junie for the past few monthsPersonal blog post, lengrand.fr (Julien's DevRel corner)

    Calls Junie one of the best coding agents tried so far, very well integrated with IntelliJ, great for Kotlin and good thanks to its test-first focus. But it is very eager to do more than asked, extremely slow (between 3 and 4 minutes on simple-ish requests), and cannot accept part of a solution.

    Published May 11, 2025self-described: Personal Kotlin side project (kbsky) in IntelliJ IDEA

    Practitioner account, self-described; direct use; counts toward Delivered outcomes, Reliability, Operational control, Cost predictability and limits How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; No tie to Junie found in GILD's check; Used it directly; Specific details; Needs recheck after 12 months

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, author profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-05-11, needs a recheck

    Counts toward

    Delivered outcomes, task completion: unfavorable, G2 (2 units before bounds)

    Counts toward

    Reliability, service availability: unfavorable, G2 (2 units before bounds)

    Counts toward

    Operational control, execution control: unfavorable, G1 (1 unit before bounds)

    Counts toward

    Operational control, models and deployment: unfavorable, G1 (1 unit before bounds)

    Counts toward

    Cost predictability and limits, cost transparency: favorable, G2 (2 units before bounds)

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Code Quality: Mixed or negative
    • Planning & Control: Mixed or negative
    • Test Writing: Positive
    • IDE Integration: Positive
    • Speed: Raised as a concern
    • Requires Strong Review: Raised as a concern
    • Stability: Raised as a concern
    • Needs Close Steering: Raised as a concern
    • Written guidelines make results better
    • How much should Junie complete before a human checkpoint?: Tighter steering

    Published on

    Engineering Blogs & Reviews

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie in IntelliJ IDEA, Code and Ask modes, Brave mode

73. Positive

    GamerTomC, Backend developer on a team at work

    Comment on: Anyone used Jetbrains Junie the coding agent? How is it?Reddit comment

    Using Junie at work on two well-structured Kotlin and Java repos, estimates their velocity rose at least 25% because it follows existing codebase patterns. Expects less value on messy codebases.

    Published May 9, 2025self-described: Team maintaining well-structured Kotlin and Java backend code

    Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-05-09, needs a recheck

    Counts toward

    Delivered outcomes, task completion: favorable, G1 (1 unit before bounds)

    Counts toward

    Delivered outcomes, context handling: favorable, G1 (1 unit before bounds), capped: C-origin

    Counts toward

    Delivered outcomes, correctness: favorable, G1 (1 unit before bounds)

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Codebase Understanding: Positive
    • Code Quality: Positive
    • Faster on self-directed work

    Published on

    Developer Communities

    Who

    Very small team

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie on two Kotlin/Java repos

74. Positive

    Snakes3727, Hacker News commenter

    Comment on: OpenAI reaches agreement to buy Windsurf for US$3BHacker News comment

    Their team compared Cursor with Junie and settled on Junie because it does a good job with Rust where Cursor did not.

    Published May 7, 2025self-described: Team working in Rust that compared Cursor and Junie

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-05-07, needs a recheck

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Code Quality: Positive
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Junie as the main tool

    Published on

    Developer Communities

    Who

    Very small team

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie, Rust codebase

75. Mixed

    mike_hearn, Hacker News commenter

    Comment on: OpenAI reaches agreement to buy Windsurf for US$3BHacker News comment

    Still prefers Claude Code for its access to more tools, since "Junie seems unable to fetch URLs", calling it a tiny gap. Finds the Junie UI quite pretty and says that with IDE index tools JetBrains "in theory" should be able to compete very strongly.

    Published May 7, 2025self-described: Uses Claude Code alongside

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-05-07, needs a recheck

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • IDE Integration: Positive
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

76. Mixed

    leonidasv, Hacker News commenter

    Comment on: OpenAI reaches agreement to buy Windsurf for US$3BHacker News comment

    Found Junie an improvement over JetBrains' earlier assistant but less smart than Cursor or Windsurf: it stops exploring too early and does not fix its own linter or compiler errors.

    Published May 7, 2025self-described: Paying JetBrains IDE user on a free Junie trial month

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-05-07, needs a recheck

    Not counted

    Delivered outcomes, context handling: repeats a stronger account in its group

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Checked Oct 30, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Codebase Understanding: Mixed or negative
    • Code Quality: Mixed or negative
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

77. Positive

    panstromek, Hacker News commenter, self-described

    Comment on: Time saved by AI offset by new work created, study suggestsHacker News comment

    After a few days with Junie, had it write long-postponed tests that were mostly correct from a one-sentence prompt. Reverts wrong attempts cheaply and finds a sweet spot of suitable tasks.

    Published May 4, 2025

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-05-04, needs a recheck

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Test Writing: Positive
    • Inconsistent Results: Discussed without concern
    • Strongest on scoped work

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

78. Positive

    ta988, Hacker News commenter, self-described

    Comment on: Show HN: Use Third Party LLM API in JetBrains AI AssistantHacker News comment

    Has used Junie a lot and finds it much better than before. It solved problems Claude Code struggled with, though not all, and is still slowish.

    Published May 3, 2025

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-05-03, needs a recheck

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Code Quality: Positive
    • Speed: Raised as a concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Junie as the main tool

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

79. Negative

    nicolaslem, Hacker News commenter

    Comment on: JetBrains defends removal of negative reviews for unpopular AI AssistantHacker News comment

    Could never get anything out of Junie in PyCharm because every query left an internal process looping at 100% CPU, so went back to Windsurf.

    Published Apr 30, 2025self-described: PyCharm user

    Practitioner account, pseudonymous; direct use; counts toward Reliability How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-04-30, needs a recheck

    Counts toward

    Reliability, run stability: unfavorable, G1 (1 unit before bounds), capped: C-relevance

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Stability: Raised as a concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie in PyCharm

80. Mixed

    bb88, Hacker News commenter

    Comment on: JetBrains defends removal of negative reviews for unpopular AI AssistantHacker News comment

    Finds Junie reasonably good and capable of fixing unit tests in Brave mode, but it struggles with large code blocks and applies changes before review, so small defined tasks work best.

    Published Apr 30, 2025self-described: Go and Python development in JetBrains IDEs

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-04-30, needs a recheck

    Not counted

    Delivered outcomes, context handling: repeats a stronger account in its group

    Not counted

    Operational control, execution control: repeats a stronger account in its group

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Codebase Understanding: Mixed or negative
    • Planning & Control: Mixed or negative
    • Test Writing: Positive
    • Requires Strong Review: Raised as a concern
    • Needs Close Steering: Raised as a concern
    • Strongest on scoped work
    • How much should Junie complete before a human checkpoint?: Tighter steering

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie, Brave mode

81. No opinion stated

    Premanand Chandrasekaran, Head of technology for the North America West at Thoughtworks

    Can vibe coding produce production-grade software?Company blog post, Thoughtworks Insights

    Chandrasekaran's team briefly tried JetBrains' Junie and Windsurf as IntelliJ IDEA plugins, finding both close to Cursor's agentic approach, but did the bulk of the work in Cursor for consistency and to minimize tool-switching.

    "We also experimented briefly with JetBrains’ Junie and Windsurf (as an IntelliJ IDEA plugin)."

    "At the time of writing, both tools offered interfaces and workflows that closely resembled Cursor’s agentic approach. However, for consistency and to minimize tool-switching overhead, we conducted the bulk of our work using Cursor."

    Published Apr 30, 2025Thoughtworks experiments building production-style software with AI agents in IntelliJ IDEA and Cursor

    Practitioner account, name and role on their organization's own page; direct use; not counted, takes no position How GILD checked it

    Origin

    Published practitioner account

    Checked

    Name and role on their organization's own page; Used it directly; Needs recheck after 12 months

    Role

    Name on a page the person does not control; employer on a page the person does not control; role on a page the person does not control (Head of technology for the North America West)

    Relationship

    Checked only from the record itself; an outside check is open

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-04-30, needs a recheck

    Not counted

    no aspect: excluded, takes no position on any assessed aspect

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

    Published on

    Engineering Blogs & Reviews

    Who

    Technical leadership

    Evidence

    Reported

    Applies to

    Needs a recheck

    Observed on

    Junie and Windsurf tried briefly as IntelliJ IDEA plugins; Cursor agent mode used for the bulk of the work

82. Negative

    thi2, Hacker News commenter

    Comment on: The hidden cost of AI codingHacker News comment

    Ran a refined prompt through Junie three times on a Vue and Kotlin Spring Boot codebase. The result was still lacking, so they discarded it and wrote the code themselves.

    Published Apr 25, 2025self-described: Employer-provided JetBrains AI; Vue frontend with Kotlin Spring Boot backend

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-04-25, needs a recheck

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Checked Oct 30, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Code Quality: Mixed or negative
    • Inconsistent Results: Raised as a concern

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie via employer JetBrains AI license

83. Positive

    Dan Vega, Developer content creator (danvega.dev)

    Beyond Auto-Complete: Junie Is Your Autonomous Coding Partner in JetBrains IDEsVideo, YouTube

    Hands-on walkthrough using Junie to explain an existing codebase and build a REST API, customized with personal guidelines, stressing that the developer stays the pilot and reviews output.

    "-step tasks autonomously, from understanding your code base to implementing new features and running tests. What makes Junie special is the ability to explore your project, generate context -appropriate code, run tests when needed, and present results for your review, all while…"

    Published Apr 20, 2025self-described: Exploring an existing codebase and building a Spring Boot CRUD REST API in IntelliJ IDEA

    Practitioner account, self-described; direct use; counts toward Operational control How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; No tie to Junie found in GILD's check; Used it directly; Specific details; Needs recheck after 12 months

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, author profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-04-20, needs a recheck

    Counts toward

    Operational control, execution control: favorable, G2 (2 units before bounds)

    Counts toward

    Operational control, models and deployment: unfavorable, G2 (2 units before bounds)

    Not counted

    Delivered outcomes, context handling: repeats a stronger account in its group

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Codebase Understanding: Positive
    • Requires Strong Review: Discussed without concern
    • Written guidelines make results better
    • Evaluated as an engineering agent

    Published on

    Videos & Talks

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie in IntelliJ IDEA

84. Positive

    shmatt, Hacker News commenter

    Comment on: JetBrains IDEs Go AI: Coding Agent, Smarter Assistance, Free TierHacker News comment

    Says Junie replaced their paid Claude Code and Cursor use: the least aggressive agent they tried, achieving about 95% of requests, but it forgot context between messages.

    Published Apr 16, 2025self-described: Web application development; previously paid for Claude Code and Cursor

    Practitioner account, pseudonymous; direct use; counts toward Reliability How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-04-16, needs a recheck

    Counts toward

    Reliability, run stability: unfavorable, G1 (1 unit before bounds), capped: C-relevance

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Not counted

    Delivered outcomes, correctness: repeats a stronger account in its group

    Checked Oct 30, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward, pending a recheck

    • Overall signal
    • Code Quality: Positive
    • Value for Money: Positive
    • Credits & Cost: Discussed without concern
    • Inconsistent Results: Discussed without concern
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Junie as the main tool

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Needs a recheck

    Observed on

    Junie EAP (free before release)

85. Positive

    Spring Framework Guru, Spring Boot course instructor

    First Look at Junie the AI Coding Agent from JetBrainsVideo, YouTube

    After a little over a week with Junie on Claude 3.7, generated and committed about 35,000 lines to a corporate repository. Finds Junie's code quality much better than GitHub Copilot Workspace's, crediting IDE hooks such as inspections and compile steps that catch missing imports.

    "…using Claude 3.7 as the LLM that it's working with. And in the last week, I'm not exaggerating, I generated and committed to uh my corporate repository for my day job about 35,000 lines of code. I've developed prompts where I can point it to…"

    Published Mar 29, 2025self-described: Spring Boot application used in the author's Udemy courses

    Practitioner account, self-described; direct use; counts toward Delivered outcomes, Reliability How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months; Specific details

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, author profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-03-29, release preview; Claude 3.7, needs a recheck

    Counts toward

    Delivered outcomes, task completion: favorable, G2 (2 units before bounds)

    Counts toward

    Delivered outcomes, correctness: mixed, G2 (2 units before bounds)

    Counts toward

    Reliability, run stability: unfavorable, G1 (1 unit before bounds)

    Not counted

    Adoption and upkeep, continuity: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • IDE Integration: Positive
    • Faster on self-directed work
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Junie as the main tool

    Published on

    Videos & Talks

    Who

    Individual practitioner

    Evidence

    Measured

    Applies to

    Still applicable

    Observed on

    Junie (early release) in IntelliJ IDEA

86. Positive

    A N M Bazlur Rahman, Java developer and conference speaker

    JetBrains Junie: My Firsthand ExperiencePersonal blog post, bazlur.ca

    Used the Junie EAP to build a website and front-end UIs outside a backend comfort zone, finishing a week of work in a day while noting it still needs oversight.

    Published Mar 3, 2025self-described: Ramadan awareness website (React) and DevNexus talk projects (Jakarta Faces, LangChain4j)

    Practitioner account, self-described; direct use; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; Junie partner: not counted as independent; Used it directly; Needs recheck after 12 months; Specific details

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    program member (self disclosed)

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-03-03, Junie EAP, needs a recheck

    Counts toward

    Delivered outcomes, correctness: unfavorable, G1 (1 unit before bounds)

    Not counted

    Delivered outcomes, task completion: excluded, praise from someone tied to the vendor, kept as an attributed position

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Test Writing: Positive
    • Inconsistent Results: Raised as a concern
    • Requires Strong Review: Discussed without concern
    • Faster on self-directed work

    Published on

    Engineering Blogs & Reviews

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie EAP

87. Positive

    Dave Leeds, Kotlin educator, typealias.com

    Junie - The New AI Coding Agent from JetBrains - Can it build a Compose Desktop app?Video, YouTube

    Lets Junie write a whole Compose Desktop app. First impressions are quite positive with pretty good code, but it feels slow and uses stale library versions.

    "…adding the guidelines - each step along the way is given its own commit, so that's great. So overall, my first impressions with Junie are quite positive. I haven't used AI for coding very much, but so far Junie seems very capable, and I'm looking…"

    Published Feb 26, 2025self-described: Compose Desktop YouTube channel browser built entirely by Junie (EAP)

    Practitioner account, self-described; structured test; counts toward Delivered outcomes How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; Junie partner: not counted as independent; Ran a structured test; Specific details; Needs recheck after 12 months

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    program member (self disclosed)

    How they know

    Ran a structured test

    Support

    1 artifact (repository)

    Date and version

    2025-02-26, Early Access Preview, needs a recheck

    Counts toward

    Delivered outcomes, correctness: mixed, G2 (2 units before bounds)

    Not counted

    Delivered outcomes, task completion: excluded, praise from someone tied to the vendor, kept as an attributed position

    Not counted

    Operational control, execution control: excluded, praise from someone tied to the vendor, kept as an attributed position

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Code Quality: Positive
    • Speed: Raised as a concern
    • Written guidelines make results better
    • Evaluated as an engineering agent

    Published on

    Videos & Talks

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie EAP in IntelliJ IDEA

88. Positive

    Duncan (Pairing with Duncan), Developer and author of Java to Kotlin, A Refactoring Guidebook

    First Look at JetBrains Junie Autonomous AI AgentVideo, YouTube

    Pairs with Junie under strict TDD. It is overly eager and needed chat and IDE restarts in the EAP, but they call it a really promising start and likes the guidelines file.

    "…so over all a really promising start and I must say that I do particularly like this guidelines file as a way of communicating in language between the AI and human and vice versa whether you use cursor or Genie if you don't already then…"

    Published Feb 17, 2025self-described: Kotlin kata project with strict TDD in IntelliJ IDEA (Junie EAP)

    Practitioner account, self-described; structured test; counts toward Reliability, Operational control, Adoption and upkeep How GILD checked it

    Origin

    Published practitioner account

    Checked

    Self-described; No tie to Junie found in GILD's check; Ran a structured test; Specific details; Needs recheck after 12 months

    Role

    Name self-described; employer unknown; role unknown

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, author profile); that is not proof none exists

    How they know

    Ran a structured test

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-02-17, early access preview, needs a recheck

    Counts toward

    Reliability, run stability: unfavorable, G1 (1 unit before bounds), capped: C-relevance

    Counts toward

    Operational control, execution control: unfavorable, G2 (2 units before bounds)

    Counts toward

    Adoption and upkeep, setup: unfavorable, G2 (2 units before bounds)

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Checked Oct 31, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Planning & Control: Mixed or negative
    • Stability: Raised as a concern
    • Written guidelines make results better

    Published on

    Videos & Talks

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie EAP in IntelliJ IDEA, Brave Mode

89. Negative

    surrTurr, Hacker News commenter

    Comment on: JetBrains Fleet drops support for Kotlin MultiplatformHacker News comment

    Tested Junie in the invite-only beta and found it very basic, roughly 30% as good as Cursor's agent mode, while noting it was still early.

    Published Feb 11, 2025self-described: Tested Junie in the invite-only beta

    Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account How GILD checked it

    Origin

    Published practitioner account

    Checked

    Pseudonymous; No tie to Junie found in GILD's check; Used it directly; Needs recheck after 12 months

    Role

    Pseudonymous account; identity not pursued

    Relationship

    No tie to Junie found in GILD's check (byline, content disclosure, platform api profile); that is not proof none exists

    How they know

    Used it directly

    Support

    No artifact or independent corroboration recorded

    Date and version

    2025-02-11, invite only beta, needs a recheck

    Not counted

    Delivered outcomes, task completion: repeats a stronger account in its group

    Checked Oct 29, 2026 under gild-policy/1.0.0. The scorecard

    Counts toward

    • Overall signal
    • Code Quality: Mixed or negative
    • Among practitioners who use Junie and another agent, which works better as the main tool?: Another agent as the main tool

    Published on

    Developer Communities

    Who

    Individual practitioner

    Evidence

    Reported

    Applies to

    Still applicable

    Observed on

    Junie invite-only beta