What engineering leaders and their teams report about Kiro, with every finding tied to its evidence.
Read the findingsPractitioner sentiment, all practitioners
+10 Mixed from 41 sources that take a side
How this figure is madeFrom 41 authors in 35 discussions. Use a task pilot to measure accepted work and review time.
Jul 2025 to Oct 2026.
73 sources from practitioners, by month and sentiment.
Across the 41 independent public accounts that give an opinion, the read is mixed and leans positive: 19 positive, 7 mixed, and 15 negative. The spec workflow draws the most praise, and all 9 sources that discuss review raise it as a concern. Engineers split 4 to 6 on whether specs repay their cost.
Positive, mixed and negative sources
The score counts each source once, as the split does. See How the numbers work.
See the evidenceBelow: 4 praise themes, 5 concerns, 1 close to agreement and 2 disagreements.
Counted for All practitioners, from 73 sources. Change group
Every capability practitioners describe positively, ranked by how many sources say so. Open one to read who said it.
Counted for All practitioners.
Positive, 9
Positive
NTRIXLM, Hacker News commenter
Comment on: Show HN: I Made a Claude Skill for Spec-Driven Development (SDD)Hacker News comment
Prefers Kiro for its spec management, but the employer provides only Claude and a personal plan is not affordable, so they rebuilt Kiro's spec flow as a Claude skill.
Practitioner account, pseudonymous; direct use; not counted, takes no position
Positive
Geshan Manandhar, Geshan's Blog
Using Spec Driven Development with AWS Kiro to add the last updated date on Eleventy blogPersonal blog post
Added a last-updated date to an Eleventy blog through a Kiro spec. Kiro wrote the requirements and design docs and a task list of up to 6 tasks, installed fast-check, added a test framework and ran tests, and the feature worked after a couple of minor changes by the author.
Practitioner account, self-described; direct use; counts toward Delivered outcomes, Operational control
Positive
wenc, Hacker News commenter
Comment on: After outages, Amazon to make senior engineers sign off on AI-assisted changesHacker News comment
Uses Kiro IDE mainly to write and pressure-test specs, often for hours or days, then hands the task list to headless agents including Kiro CLI and Claude Code.
Practitioner account, pseudonymous; direct use; counts toward Reliability
Mixed
achenatx, Hacker News commenter, self-described
Comment on: How I use Claude Code: Separation of planning and executionHacker News comment
The spec flow lets Kiro run an hour or more on small chunks with mostly correct results that still need fixes. Steering files use up context and force resets.
Practitioner account, pseudonymous; direct use; counts toward Reliability
Positive
nikolay, Hacker News commenter, self-described
Comment on: Is vibe coding dying?Hacker News comment
Gets impressive results from Kiro's spec-based approach, though it takes much longer than generating a program from one prompt.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Positive
Semaphor, Hacker News commenter, self-described
Comment on: Developing with Kiro: Amazon's New Agentic IDEHacker News comment
After using it for a while, other tools feel lacking: the multistep spec process keeps the model in line, and they left Kiro only when Sonnet 4 was overloaded.
Practitioner account, pseudonymous; direct use; counts toward Reliability
Positive
danudey, Hacker News commenter
Comment on: Cognition (Devin AI) to Acquire WindsurfHacker News comment
Where Copilot and Cursor had produced broken code, Kiro wrote requirements, a design, and a traced task list, then working Rust code for a project the author barely knew.
Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes
Positive
Engineering manager, cloud infrastructure platform team
Kiro showcase: Automating Changes Across Several Repos with Spec-Driven Development and Custom Sub-AgentsPersonal blog post, DEV Community
De Paolis used a Kiro spec and an agent run as parallel kiro-cli subagents to change 30+ repos, spending "probably less than 90 minutes" and roughly 80 credits of a 1K monthly allowance on the US$20/month plan, against "likely 4-6 hours" of work by hand.
Practitioner account, self-described; direct use; counts toward Delivered outcomes, Reliability, Operational control, Cost predictability and limits, Adoption and upkeep
Mixed
lan321, Hacker News commenter
Comment on: Gemini 3Hacker News comment
Likes how Kiro splits large projects into bite-sized chunks, but could not get it to implement solitaire correctly.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Mixed or negative, 2
Negative
ChilledTonic, Hacker News commenter, self-described
Comment on: Kiro Pricing Plans Are Now LiveHacker News comment
Likes the idea of spec documents and steering, but found them underwhelming in use: they often confused the model more than they produced working code.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Mixed
Birgitta Böckeler, Distinguished Engineer at Thoughtworks
Understanding Spec-Driven-Development: Kiro, spec-kit, and TesslEngineering article, martinfowler.com
Böckeler found Kiro the simplest, most lightweight of three spec tools. Asked to fix a small bug, it produced 4 user stories with 16 acceptance criteria, which the author likened to a sledgehammer cracking a nut, preferring to review code over all those markdown files.
Practitioner account, self-described; direct use; counts toward Delivered outcomes
Counted for All practitioners.
Positive, 7
Positive
Engineering manager, cloud infrastructure platform team
Kiro showcase: Automating Changes Across Several Repos with Spec-Driven Development and Custom Sub-AgentsPersonal blog post, DEV Community
De Paolis used a Kiro spec and an agent run as parallel kiro-cli subagents to change 30+ repos, spending "probably less than 90 minutes" and roughly 80 credits of a 1K monthly allowance on the US$20/month plan, against "likely 4-6 hours" of work by hand.
Practitioner account, self-described; direct use; counts toward Delivered outcomes, Reliability, Operational control, Cost predictability and limits, Adoption and upkeep
Positive
Geshan Manandhar, Geshan's Blog
Using Spec Driven Development with AWS Kiro to add the last updated date on Eleventy blogPersonal blog post
Added a last-updated date to an Eleventy blog through a Kiro spec. Kiro wrote the requirements and design docs and a task list of up to 6 tasks, installed fast-check, added a test framework and ran tests, and the feature worked after a couple of minor changes by the author.
Practitioner account, self-described; direct use; counts toward Delivered outcomes, Operational control
Mixed
achenatx, Hacker News commenter, self-described
Comment on: How I use Claude Code: Separation of planning and executionHacker News comment
The spec flow lets Kiro run an hour or more on small chunks with mostly correct results that still need fixes. Steering files use up context and force resets.
Practitioner account, pseudonymous; direct use; counts toward Reliability
Mixed
hmokiguess, Hacker News commenter
Comment on: How do you pick a Coding Agent HN?Hacker News comment
Tried Kiro on AWS credits and found it outperformed Claude Code in some cases, but not by enough to switch from Claude Code.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Positive
sqrtc, Hacker News commenter, self-described
Comment on: Developing with Kiro: Amazon's New Agentic IDEHacker News comment
On a hard date-time parsing problem, the spec flow forced careful thinking about the problem, and watching Kiro work the task list was impressive. Terminal runs still needed approval one by one.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Mixed
Cu3PO42, Hacker News commenter
Comment on: Developing with Kiro: Amazon's New Agentic IDEHacker News comment
From a short spec, Kiro wrote about 5,000 lines with tests that worked, but the tool was over-engineered, and the author cut it to about 800 lines by hand.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Positive
danudey, Hacker News commenter
Comment on: Cognition (Devin AI) to Acquire WindsurfHacker News comment
Where Copilot and Cursor had produced broken code, Kiro wrote requirements, a design, and a traced task list, then working Rust code for a project the author barely knew.
Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes
Mixed or negative, 7
Negative
elfly, Hacker News commenter, self-described
Comment on: Anthropic blocks third-party use of Claude Code subscriptionsHacker News comment
Kiro starts well with planning but is hard to control: it changes files on a whim and switches opinions mid-answer.
Practitioner account, pseudonymous; direct use; counts toward Reliability
Negative
falkensmaize, Hacker News commenter
Comment on: Opus 4.5 is not the normal AI agent experience that I have had thus farHacker News comment
Using Opus 4.5 through Kiro at work, it could not upgrade Nx and yarn in a 20-app monorepo, invented config options, and the engineer fixed it by hand.
Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes
Mixed
lan321, Hacker News commenter
Comment on: Gemini 3Hacker News comment
Likes how Kiro splits large projects into bite-sized chunks, but could not get it to implement solitaire correctly.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Negative
jtrn, Hacker News commenter, self-described
Comment on: Cursor 1.7Hacker News comment
Asked to build a simple MCP server from documentation, Kiro planned and executed impressively for six hours without reaching a basic MVP.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Negative
sgt101, Hacker News commenter, self-described
Comment on: Developing with Kiro: Amazon's New Agentic IDEHacker News comment
Kiro spent half a day writing code and tests for an app the author could not use, while Replit plus two hours in Cursor produced a working result.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Negative
codeduck, Hacker News commenter, self-described
Comment on: Meta Security Researcher's AI Agent Accidentally Deleted Her EmailsHacker News comment
Spent an hour persuading Kiro that a failing local MCP call came from its own missing Authorization header; it was not attributed to the proxy. Only the logs convinced it.
Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes
Negative
k3vinw, Hacker News commenter
Comment on: Uncle Bob: It's OverHacker News comment
During a refactor, Kiro blamed the testing library. The engineer stepped in and found the real cause, an async bug hidden by faulty test doubles.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Counted for All practitioners.
Positive, 6
Positive
epistasis, Hacker News commenter, self-described
Comment on: Codex logging bug may write TBs to local SSDsHacker News comment
Kiro is surprisingly good. With more reliable session save and resume, and remote sessions, they would switch to it full time.
Practitioner account, pseudonymous; direct use; counts toward Reliability
Positive
marysol5, Hacker News commenter
Comment on: Alphabet announces $80B equity capital raise to expand AI infra and computeHacker News comment
Kiro, which they use at work, prompts the user when a command needs input such as a sudo password, while another agent sat thinking until they intervened.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Positive
avereveard, Hacker News commenter
Comment on: 1M context is now generally available for Opus 4.6 and Sonnet 4.6Hacker News comment
Uses Kiro at work and prefers in-IDE agents because reviewing changes and pulling in context is faster. Task delegation is the feature that matters to them.
Practitioner account, pseudonymous; direct use; not counted, takes no position
Positive
mlaretallack, Hacker News commenter
Comment on: Ask HN: Are you using an agent orchestrator to write code?Hacker News comment
Runs a separate Kiro instance per project in its own desktop workspace, about four at a time, and relies on Kiro's notifications when an agent needs input.
Practitioner account, pseudonymous; direct use; not counted, takes no position
Positive
danny_codes, Hacker News commenter, self-described
Comment on: GLM 5.2 and the coming AI margin collapseHacker News comment
Switched to Kiro and finds it essentially the same as their previous tool, slightly better for showing what the harness is doing.
Practitioner account, pseudonymous; direct use; not counted, takes no position
Positive
cobolcomesback, Hacker News commenter, self-described
Comment on: Anthropic blocks third-party use of Claude Code subscriptionsHacker News comment
Finds Kiro better than Claude Code: result quality is similar, but Claude Code got stuck on tool-call errors where Kiro just worked, and they prefer Kiro's simpler UX.
Practitioner account, pseudonymous; direct use; counts toward Reliability, Operational control
Mixed or negative, 3
Negative
s4i, Hacker News commenter, self-described
Comment on: Our decision on Cursor following its acquisition by SpaceXHacker News comment
Looking for a Cursor replacement, tried Kiro, which shares the philosophy but lacks most of the Cursor features they rely on.
Practitioner account, pseudonymous; direct use; not counted, takes no position
Negative
sirtaj, Hacker News commenter
Comment on: Anthropic blocks third-party use of Claude Code subscriptionsHacker News comment
Uses Kiro CLI with Opus at work and finds it clearly inferior to Claude Code within the first 30 seconds.
Practitioner account, pseudonymous; direct use; not counted, takes no position
Mixed
willsmith72, Hacker News commenter, self-described
Comment on: Kiro: A new agentic IDEHacker News comment
First impressions: autopilot should be off by default as in Claude Code and Cline, the Kiro panel ignores the user's sidebar layout, and the chat box is busy.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Counted for All practitioners.
Positive, 2
Positive
Geshan Manandhar, Geshan's Blog
Using Spec Driven Development with AWS Kiro to add the last updated date on Eleventy blogPersonal blog post
Added a last-updated date to an Eleventy blog through a Kiro spec. Kiro wrote the requirements and design docs and a task list of up to 6 tasks, installed fast-check, added a test framework and ran tests, and the feature worked after a couple of minor changes by the author.
Practitioner account, self-described; direct use; counts toward Delivered outcomes, Operational control
Positive
danudey, Hacker News commenter
Comment on: Cognition (Devin AI) to Acquire WindsurfHacker News comment
Where Copilot and Cursor had produced broken code, Kiro wrote requirements, a design, and a traced task list, then working Rust code for a project the author barely knew.
Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes
Mixed or negative, 5
Mixed
achenatx, Hacker News commenter, self-described
Comment on: How I use Claude Code: Separation of planning and executionHacker News comment
The spec flow lets Kiro run an hour or more on small chunks with mostly correct results that still need fixes. Steering files use up context and force resets.
Practitioner account, pseudonymous; direct use; counts toward Reliability
Negative
thegagne, Hacker News commenter
Comment on: How Google got its groove back and edged ahead of OpenAIHacker News comment
A spec run produced odd requirements, an over-complex design, a long task list, and many passing tests but no useful code, after 8 hours and 200+ credits.
Practitioner account, pseudonymous; direct use; counts toward Operational control
Negative
KaiMagnus, Hacker News commenter, self-described
Comment on: Getting AI to work in complex codebasesHacker News comment
The requirements, design, and task documents caught edge cases at first, but the planning still produced mediocre code that was mostly thrown away. Small fast iterations work better for this author.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Mixed
Cu3PO42, Hacker News commenter
Comment on: Developing with Kiro: Amazon's New Agentic IDEHacker News comment
From a short spec, Kiro wrote about 5,000 lines with tests that worked, but the tool was over-engineered, and the author cut it to about 800 lines by hand.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Negative
sgt101, Hacker News commenter, self-described
Comment on: Developing with Kiro: Amazon's New Agentic IDEHacker News comment
Kiro spent half a day writing code and tests for an app the author could not use, while Replit plus two hours in Cursor produced a working result.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Every concern practitioners raise, most raised first.
Counted for All practitioners.
Negative
codeduck, Hacker News commenter, self-described
Comment on: Meta Security Researcher's AI Agent Accidentally Deleted Her EmailsHacker News comment
Spent an hour persuading Kiro that a failing local MCP call came from its own missing Authorization header; it was not attributed to the proxy. Only the logs convinced it.
Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes
Negative
k3vinw, Hacker News commenter
Comment on: Uncle Bob: It's OverHacker News comment
During a refactor, Kiro blamed the testing library. The engineer stepped in and found the real cause, an async bug hidden by faulty test doubles.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Mixed
Cu3PO42, Hacker News commenter
Comment on: Developing with Kiro: Amazon's New Agentic IDEHacker News comment
From a short spec, Kiro wrote about 5,000 lines with tests that worked, but the tool was over-engineered, and the author cut it to about 800 lines by hand.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Positive
Engineering manager, cloud infrastructure platform team
Kiro showcase: Automating Changes Across Several Repos with Spec-Driven Development and Custom Sub-AgentsPersonal blog post, DEV Community
De Paolis used a Kiro spec and an agent run as parallel kiro-cli subagents to change 30+ repos, spending "probably less than 90 minutes" and roughly 80 credits of a 1K monthly allowance on the US$20/month plan, against "likely 4-6 hours" of work by hand.
Practitioner account, self-described; direct use; counts toward Delivered outcomes, Reliability, Operational control, Cost predictability and limits, Adoption and upkeep
Mixed
achenatx, Hacker News commenter, self-described
Comment on: How I use Claude Code: Separation of planning and executionHacker News comment
The spec flow lets Kiro run an hour or more on small chunks with mostly correct results that still need fixes. Steering files use up context and force resets.
Practitioner account, pseudonymous; direct use; counts toward Reliability
Negative
elfly, Hacker News commenter, self-described
Comment on: Anthropic blocks third-party use of Claude Code subscriptionsHacker News comment
Kiro starts well with planning but is hard to control: it changes files on a whim and switches opinions mid-answer.
Practitioner account, pseudonymous; direct use; counts toward Reliability
Negative
thegagne, Hacker News commenter
Comment on: How Google got its groove back and edged ahead of OpenAIHacker News comment
A spec run produced odd requirements, an over-complex design, a long task list, and many passing tests but no useful code, after 8 hours and 200+ credits.
Practitioner account, pseudonymous; direct use; counts toward Operational control
Negative
falkensmaize, Hacker News commenter
Comment on: Opus 4.5 is not the normal AI agent experience that I have had thus farHacker News comment
Using Opus 4.5 through Kiro at work, it could not upgrade Nx and yarn in a 20-app monorepo, invented config options, and the engineer fixed it by hand.
Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes
Negative
KaiMagnus, Hacker News commenter, self-described
Comment on: Getting AI to work in complex codebasesHacker News comment
The requirements, design, and task documents caught edge cases at first, but the planning still produced mediocre code that was mostly thrown away. Small fast iterations work better for this author.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
GILD's fact checks, which don't count as sources
Checked Oct 12, 2026.Kiro IDE creates a checkpoint with each prompt, and restoring one reverts the agent's file changes and context. Changes made by shell commands or MCP tools are not tracked. Human review before merge remains the team's job. Checkpoints and rewind
Open them in the recordCounted for All practitioners.
Raised as a concern, 9
Negative
rsammons, GitHub user
Reliability regression: file edits fail, IDE freezes, spec operations die with network errors, and in-app bug reporter is brokenGitHub issue on kirodotdev/Kiro
Design generation from an existing spec failed three times with a backend network error after about 20 minutes. Each attempt re-ran 25 or more tool calls, and credits were still consumed for the failed work.
"~20 minutes elapsed before any problem was surfaced; three identical failures; no design document produced; credits consumed for failed operations…"
Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes, Reliability, Cost predictability and limits
Negative
rgoltz, GitHub user
Kiro IDE - Extension host crashes when processing large LLM responses (event loop blocked)GitHub issue on kirodotdev/Kiro
The Kiro agent crashed the editor's extension host 3 times in one working day across 2 instances, losing in-progress work each time and making sessions longer than 30 minutes unusable.
"This happens repeatedly and reliably, 3 crashes in a single working day across 2 Kiro instances."
Practitioner account, pseudonymous; direct use; counts toward Reliability
Mixed
Developer
When a Kiro CLI Update Silently Breaks Your MCP Servers: The Approved Environment Variables GotchaPersonal blog post, DEV Community
A Kiro CLI update stopped expanding unapproved environment variables in MCP configs, so the author's MCP server began returning 401 errors overnight until the variables were approved.
Practitioner account, self-described; direct use; counts toward Reliability, Operational control
No opinion stated
imtaxu, GitHub user and AWS Organizations administrator
Kiro subscription assignment fails on Organizations management account and routes to standalone CreateAssignment pathGitHub issue on kirodotdev/Kiro
Could not assign any paid Kiro plan to Identity Center users or groups from an AWS Organizations management account. CloudTrail showed the call going to a standalone Amazon Q Developer subscription path and failing with AccessDenied.
"I cannot assign a paid Kiro subscription to a user or group from AWS IAM Identity Center."
Practitioner account, pseudonymous; ran the rollout; counts toward Adoption and upkeep
Negative
huey77, Hacker News commenter
Comment on: AWS Kiro accounts suspended en masse - AWS flagging fraud detection issuesHacker News comment
Their Kiro account was suspended after an upgrade to the US$200 Power plan, with no explanation, and they link reports suggesting it was widespread.
not counted, repeats a stronger account
Mixed
ActorNightly, Hacker News commenter, self-described
Comment on: Kiro: A new agentic IDEHacker News comment
On launch day, liked the product but found Kiro sitting idle with its helper plugin using heavy CPU for no reason.
Practitioner account, pseudonymous; direct use; counts toward Reliability
Positive
epistasis, Hacker News commenter, self-described
Comment on: Codex logging bug may write TBs to local SSDsHacker News comment
Kiro is surprisingly good. With more reliable session save and resume, and remote sessions, they would switch to it full time.
Practitioner account, pseudonymous; direct use; counts toward Reliability
Mixed
smarchypants, GitHub user
New sessions immediately hit 85%+ context usage and enter summarization loop, even with minimal steering docs and simple promptsGitHub issue on kirodotdev/Kiro
After 6 months of normal use, every new session hit 85% context on the first exchange because Kiro ignored .kiroignore and loaded all 10,080 workspace files. Only moving the files out of the workspace fixed it.
"Actual behavior: All 10,080 files were still loaded into context (100,000+ tokens)…"
Practitioner account, pseudonymous; structured test; counts toward Reliability
Negative
hatmanstack, Hacker News commenter, self-described
Comment on: Understanding Spec-Driven-Development: Kiro, Spec-Kit, and TesslHacker News comment
The spec approach generated massive task lists and deleted code unpredictably without reverting. Plain iteration is simpler for this author.
Practitioner account, pseudonymous; direct use; counts toward Reliability
Discussed without concern, 1
Positive
Semaphor, Hacker News commenter, self-described
Comment on: Developing with Kiro: Amazon's New Agentic IDEHacker News comment
After using it for a while, other tools feel lacking: the multistep spec process keeps the model in line, and they left Kiro only when Sonnet 4 was overloaded.
Practitioner account, pseudonymous; direct use; counts toward Reliability
Counted for All practitioners.
Raised as a concern, 4
Negative
thegagne, Hacker News commenter
Comment on: How Google got its groove back and edged ahead of OpenAIHacker News comment
A spec run produced odd requirements, an over-complex design, a long task list, and many passing tests but no useful code, after 8 hours and 200+ credits.
Practitioner account, pseudonymous; direct use; counts toward Operational control
Negative
hatmanstack, Hacker News commenter, self-described
Comment on: Understanding Spec-Driven-Development: Kiro, Spec-Kit, and TesslHacker News comment
The spec approach generated massive task lists and deleted code unpredictably without reverting. Plain iteration is simpler for this author.
Practitioner account, pseudonymous; direct use; counts toward Reliability
Mixed
Birgitta Böckeler, Distinguished Engineer at Thoughtworks
Understanding Spec-Driven-Development: Kiro, spec-kit, and TesslEngineering article, martinfowler.com
Böckeler found Kiro the simplest, most lightweight of three spec tools. Asked to fix a small bug, it produced 4 user stories with 16 acceptance criteria, which the author likened to a sledgehammer cracking a nut, preferring to review code over all those markdown files.
Practitioner account, self-described; direct use; counts toward Delivered outcomes
Negative
CuriouslyC, Hacker News commenter
Comment on: The future of software development is software developersHacker News comment
Not a fan: Kiro is too heavyweight, and agents do not follow its detailed specs closely enough to repay the time spent writing them.
Practitioner account, pseudonymous; opinion; not counted, from a competitor's side
Discussed without concern, 1
Positive
nikolay, Hacker News commenter, self-described
Comment on: Is vibe coding dying?Hacker News comment
Gets impressive results from Kiro's spec-based approach, though it takes much longer than generating a program from one prompt.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
GILD's fact checks, which don't count as sources
Checked Oct 12, 2026.Kiro now offers Quick Spec and Bugfix Specs alongside feature specs, and runs independent tasks in parallel waves. Teams still decide which changes warrant a spec. Specs
Open them in the recordCounted for All practitioners.
Raised as a concern, 3
Negative
codeduck, Hacker News commenter, self-described
Comment on: Meta Security Researcher's AI Agent Accidentally Deleted Her EmailsHacker News comment
Spent an hour persuading Kiro that a failing local MCP call came from its own missing Authorization header; it was not attributed to the proxy. Only the logs convinced it.
Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes
Negative
elfly, Hacker News commenter, self-described
Comment on: Anthropic blocks third-party use of Claude Code subscriptionsHacker News comment
Kiro starts well with planning but is hard to control: it changes files on a whim and switches opinions mid-answer.
Practitioner account, pseudonymous; direct use; counts toward Reliability
Negative
falkensmaize, Hacker News commenter
Comment on: Opus 4.5 is not the normal AI agent experience that I have had thus farHacker News comment
Using Opus 4.5 through Kiro at work, it could not upgrade Nx and yarn in a 20-app monorepo, invented config options, and the engineer fixed it by hand.
Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes
Discussed without concern, 3
Mixed
achenatx, Hacker News commenter, self-described
Comment on: How I use Claude Code: Separation of planning and executionHacker News comment
The spec flow lets Kiro run an hour or more on small chunks with mostly correct results that still need fixes. Steering files use up context and force resets.
Practitioner account, pseudonymous; direct use; counts toward Reliability
Positive
Semaphor, Hacker News commenter, self-described
Comment on: Developing with Kiro: Amazon's New Agentic IDEHacker News comment
After using it for a while, other tools feel lacking: the multistep spec process keeps the model in line, and they left Kiro only when Sonnet 4 was overloaded.
Practitioner account, pseudonymous; direct use; counts toward Reliability
Positive
danudey, Hacker News commenter
Comment on: Cognition (Devin AI) to Acquire WindsurfHacker News comment
Where Copilot and Cursor had produced broken code, Kiro wrote requirements, a design, and a traced task list, then working Rust code for a project the author barely knew.
Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes
Counted for All practitioners.
Raised as a concern, 3
Mixed
astqs, Hacker News commenter, self-described
Comment on: AI coding at home without going brokeHacker News comment
Spent about 200 euros of usage in one weekend in December. It was money well spent, though they expect heavy use to cost far more at break-even prices.
Practitioner account, pseudonymous; direct use; not counted, takes no position
Negative
thegagne, Hacker News commenter
Comment on: How Google got its groove back and edged ahead of OpenAIHacker News comment
A spec run produced odd requirements, an over-complex design, a long task list, and many passing tests but no useful code, after 8 hours and 200+ credits.
Practitioner account, pseudonymous; direct use; counts toward Operational control
Negative
rsammons, GitHub user
Reliability regression: file edits fail, IDE freezes, spec operations die with network errors, and in-app bug reporter is brokenGitHub issue on kirodotdev/Kiro
Design generation from an existing spec failed three times with a backend network error after about 20 minutes. Each attempt re-ran 25 or more tool calls, and credits were still consumed for the failed work.
"~20 minutes elapsed before any problem was surfaced; three identical failures; no design document produced; credits consumed for failed operations…"
Practitioner account, pseudonymous; direct use; counts toward Delivered outcomes, Reliability, Cost predictability and limits
Discussed without concern, 5
Positive
forgotaccount3, Hacker News commenter
Comment on: AI sticker shock hits corporate AmericaHacker News comment
Their employer uses Kiro on a deal below public pricing, has enabled overages for everyone, and moves people between tiers by usage. They see clear productivity per dollar.
Practitioner account, pseudonymous; direct use; not counted, repeats a stronger account
Positive
Miner49er, Hacker News commenter
Comment on: I believe there are entire companies right now under AI psychosisHacker News comment
Struggles to use the 1,000 credits a US$20 plan gives each month, despite using Kiro more than anyone else on the team.
Practitioner account, pseudonymous; direct use; counts toward Cost predictability and limits
Positive
topce, Hacker News commenter
Comment on: Microsoft and OpenAI end their exclusive and revenue-sharing dealHacker News comment
Having used both Copilot and Kiro, lists Sonnet at 1 and Opus at 3 on Copilot against 1.3 and 2.2 on Kiro, and says "IMHO lot of people will switch to kiro and or deep seek", adding "it look like AWS done best inference".
Practitioner account, pseudonymous; direct use; counts toward Operational control
Positive
NTRIXLM, Hacker News commenter
Comment on: Show HN: I Made a Claude Skill for Spec-Driven Development (SDD)Hacker News comment
Prefers Kiro for its spec management, but the employer provides only Claude and a personal plan is not affordable, so they rebuilt Kiro's spec flow as a Claude skill.
Practitioner account, pseudonymous; direct use; not counted, takes no position
Positive
Engineering manager, cloud infrastructure platform team
Kiro showcase: Automating Changes Across Several Repos with Spec-Driven Development and Custom Sub-AgentsPersonal blog post, DEV Community
De Paolis used a Kiro spec and an agent run as parallel kiro-cli subagents to change 30+ repos, spending "probably less than 90 minutes" and roughly 80 credits of a 1K monthly allowance on the US$20/month plan, against "likely 4-6 hours" of work by hand.
Practitioner account, self-described; direct use; counts toward Delivered outcomes, Reliability, Operational control, Cost predictability and limits, Adoption and upkeep
GILD's fact checks, which don't count as sources
Checked Oct 12, 2026.Since September 2025, every prompt, spec task, and hook draws on one monthly credit pool, metered to 0.01 credit and scaled by the model's multiplier. Paid plans can buy add-on credits at US$0.04 each. Reports about vibe and spec requests describe a system that no longer applies. Kiro pricing
Open them in the recordStatements this group's sources agree on, then the ones close to agreement.
Close to agreement: each needs 15 supporting sources
Writing requirements and a design before code surfaces gaps early and makes the agent's work more predictable.
5 supporting sourcesQuestions the sources split on, with the count on each side.
Evidence remains mixed
Made Kiro the main tool, 5 sources
Some moved their IDE to Kiro, added it beside Copilot, or prefer it to Claude Code for its simpler interface and fewer stuck runs.
Kept another tool, 7 sources
Others use Kiro only to write specs and implement in Claude Code, or stayed with Cursor, Copilot, or Claude Code.
Evidence remains mixed
Specs pay off, 4 sources
Some report fewer rework cycles, exposed assumptions, and a working feature after a few minor fixes once the spec is right.
Specs cost more than they return, 6 sources
Others find specs heavy for small changes, ignored by the agent, or producing over-built code they throw away. One small bug became 4 user stories with 16 acceptance criteria.
The three groups, then every practitioner by the context they work in, and how emphasis changes with scale.
Three groups drawn from the same sources. Choose a group to select whose experience counts. Within that group, each author counts once in the practitioner sentiment, with their latest view. Signals open on All practitioners. Leaders and teams have 2 accounts; GILD prints their sentiment from 20.
11 sources. These accounts focus on rollout decisions, shared steering and hooks, review load, and what changes when Kiro moves from one engineer to a team standard.
62 sources. These accounts focus on the spec workflow, agent reliability, credit use, and how Kiro compares with the editor or agent the author already uses.
73 sources. This gives the broadest read, but combines individual and team contexts that can reach different conclusions.
73 sources from practitioners, by the context they work in. Filled rows count for this group.
Emphasizes Spec workflow, reliability, and credits
Emphasizes Agent help on a shared project, and plan value
Emphasizes Spec-first planning and multi-repository changes
Emphasizes Required use, team skills, and company-wide overages
What changes with scale. Individual builders write most about the spec workflow and whether it repays its overhead. The few team accounts describe the shared setup around it: team skills and AGENTS.md files inside AWS, custom subagents for changes across many repositories, and overages switched on for a whole company.
Where this group's sources were published.
Counted for All practitioners, from 73 sources. Change group
89 sources for this group, 16 of them documentation
Older evidence is retained when it still applies. Sources that depend on early-access behavior, an older model list, or request-based pricing stay in the history, marked Needs a recheck or Out of date.
See sources by originWho each source is, how it knows, whether it still applies and what it counts toward.
Newest first, on its own page. Every count above opens its sources there.
Read all 89 sourcesAbout this read
A running review of what engineering teams report about Kiro, from individual builders to company-wide rollouts. Each finding links to its sources, and each conflict shows both sides.
Counts cover only the sources GILD reviewed.
Choose a group to select whose experience counts. Within that group, each author counts once in the practitioner sentiment, with their latest view. GILD files each source in a group from how the account describes its author. A role or employer marked self-described is what the source states about themselves. Documentation checks facts for every group and never counts toward sentiment.
A source's date and whether it still applies are separate: a recent post can describe an old version, and an old one can still hold.
Flagged for a recheck, of the sources that take a side: leaders and teams 0 of 2, individual builders 8 of 39, all practitioners 8 of 41.
Each row opens its sources for the group you are reading.
| Finding | Engineering leaders and teams | Individual builders | All practitioners | Sources |
|---|---|---|---|---|
| Current readCurrent sentiment | 2 positive, 0 mixed, 0 negative | 17 positive, 7 mixed, 15 negative | 19 positive, 7 mixed, 15 negative | Sources |
| PraiseSpec Workflow | 1 of 1 positive (under the minimum) | 8 of 10 positive | 9 of 11 positive | Sources |
| PraiseAgent Execution | 1 of 1 positive (under the minimum) | 6 of 13 positive | 7 of 14 positive | Sources |
| PraiseEditor & CLI Experience | 0 of 0 positive (under the minimum) | 6 of 9 positive | 6 of 9 positive | Sources |
| PraiseSteering, Hooks & Agents | 0 of 0 positive (under the minimum) | 2 of 3 positive (under the minimum) | 2 of 3 positive (under the minimum) | Sources |
| PraiseCode Quality | 0 of 0 positive (under the minimum) | 2 of 7 positive | 2 of 7 positive | Sources |
| ConcernRequires Strong Review | 1 of 1 raise it (under the minimum) | 8 of 8 raise it | 9 of 9 raise it | Sources |
| ConcernAgent & Service Reliability | 1 of 1 raise it (under the minimum) | 8 of 9 raise it | 9 of 10 raise it | Sources |
| ConcernCredits / Cost | 0 of 3 raise it (under the minimum) | 3 of 5 raise it | 3 of 8 raise it | Sources |
| ConcernSpec Overhead | 0 of 0 raise it (under the minimum) | 4 of 5 raise it | 4 of 5 raise it | Sources |
| ConcernContext Management | 0 of 0 raise it (under the minimum) | 2 of 2 raise it (under the minimum) | 2 of 2 raise it (under the minimum) | Sources |
| ConcernNeeds Close Steering | 0 of 0 raise it (under the minimum) | 3 of 6 raise it | 3 of 6 raise it | Sources |
| ConsensusPlanning first improves the result | 0 supporting (under the minimum) | 5 supporting (below threshold) | 5 supporting (below threshold) | Sources |
| DisagreementAfter trying Kiro next to other tools, do practitioners make it their main tool? | Made Kiro the main tool 0, Kept another tool 0 (under the minimum) | Made Kiro the main tool 5, Kept another tool 7 | Made Kiro the main tool 5, Kept another tool 7 | Sources |
| DisagreementDo Kiro's specs pay for the time they take? | Specs pay off 0, Specs cost more than they return 0 (under the minimum) | Specs pay off 4, Specs cost more than they return 6 | Specs pay off 4, Specs cost more than they return 6 | Sources |
| Current readEvidence mix | 27 sources | 78 sources | 89 sources | Sources |
GILD's checks of claims against official documentation, kept apart from the source counts.
