What engineering leaders and their teams report about Qodo, with every finding tied to its evidence.
Read the findingsPractitioner sentiment, leaders and teams
+17 Mixed from 23 sources that take a side
How this figure is madeFrom 23 authors in 22 discussions. Small sample. Run a pilot on live pull requests to measure which comments engineers act on and which they dismiss.
Jan 2025 to Oct 2026.
53 sources from practitioners, by month and sentiment.
Of 23 team and maintainer accounts that give an opinion, 9 speak well of Qodo, 9 report both gains and problems, and 5 speak against it. All 7 team sources that discuss bug catches report real ones, and comment noise is raised by 8 of the 10 that discuss it. At Beko, developers resolved 73.8% of the reviewer's comments while average pull request closure time rose from 5 hours 52 minutes to 8 hours 20 minutes. Of the 8 teams that weighed keeping Qodo, 5 kept it and 3 dropped it.
Positive, mixed and negative sources
The score counts each source once, as the split does. See How the numbers work.
See the evidenceBelow: 2 praise themes, 2 concerns, 1 point of agreement and 1 disagreement.
Counted for Engineering leaders and teams, from 24 sources. Change group
Every capability practitioners describe positively, ranked by how many sources say so. Open one to read who said it.
Counted for Engineering leaders and teams.
Positive
Vivaan K., reviewer on G2
Big-Picture Code Reviews with Qodo, Great for Catching Logic and Dependency IssuesG2 review, syndicated on AWS Marketplace
Vivaan K. rates Qodo 4 of 5. The data engineer values repository-wide checks for logic and dependency issues before merge. Judging findings still requires pipeline and business knowledge.
Positive
Subhramit Basu Bhowmick, maintainer at JabRef
Add Qodo acknowledgement actionGitHub pull request on JabRef/jabref
JabRef maintainers say Qodo has caught things they would otherwise have overlooked, and they kept having to remind authors to read its comments. They added a check that fails until the author replies to each Qodo comment or a push makes it outdated.
"Over time, Qodo has proved to be useful in catching things that would've otherwise been overlooked."
Positive
Mohit Kumar, maintainer of cross-cluster-replication at OpenSearch
[Enhancement] Enable AI-powered PR code review and suggestions (PR-Agent via opensearch-build)GitHub issue on opensearch-project/cross-cluster-replication
A maintainer asked to bring the OpenSearch project's PR-Agent reviewer on Amazon Bedrock to cross-cluster-replication, saying that on the main OpenSearch repository it catches null pointer risks, wrong allocation IDs, missing error handling, and design problems before human review. The project onboarded the workflow to the repository.
"…it catches issues like null pointer risks, incorrect allocation ID usage, missing error handling, and architectural concerns before human reviewers even look at the PR…"
Mixed
Umut Cihan, Bilkent University, with co-authors from Beko
Automated Code Review In PracticeConference talk page, ICSE 2025 Software Engineering in Practice track
At Beko, developers resolved 73.8% of the comments from a reviewer built on open source Qodo PR-Agent. Average pull request closure time rose from 5 hours 52 minutes to 8 hours 20 minutes. Developers credited faster bug detection and reported faulty reviews and irrelevant comments.
"%73.8 of automated code review comments were labeled as resolved. However, the overall average pull request closure duration increased from five hours 52 minutes to eight hours 20 minutes…"
Positive
Jim Freeman, founder at Punt Labs
ci(review): pilot Qodo-first review with bounded rounds (vox-2q95)GitHub pull request on punt-labs/vox
Across 923 pull requests, Punt Labs hand-classified 50 comments per bot. Of Qodo's, 20% were real bugs and 54% were nits, with 12% false positives, and Qodo caused the fewest extra pushes. The team turned Copilot review off and made Qodo the first reviewer, running once on open with only high-severity findings posted inline.
"Copilot is the noise source; Bugbot finds the most real bugs; Qodo converges but its false positives were all process "Rule violations"."
Positive
Ambrish Vadnerkar, Medium
How I Added AI Code Reviews to Our GitLab Pipeline Using PR-AgentPersonal blog post, Medium
Ambrish Vadnerkar found generated descriptions the biggest benefit after a few months of PR-Agent on every merge request. It caught a hardcoded API key, but noise persisted. Vadnerkar worries that the team will rely on it too much.
Positive
outofdistro, Hacker News commenter
Comment on: Graphite is joining CursorHacker News comment
outofdistro's team has used Qodo for a while. It occasionally finds a serious issue, but effort labels and file-by-file walkthroughs help most by speeding human review.
Counted for Engineering leaders and teams.
Positive, 3
Positive
Jim Freeman, founder at Punt Labs
ci(review): pilot Qodo-first review with bounded rounds (vox-2q95)GitHub pull request on punt-labs/vox
Across 923 pull requests, Punt Labs hand-classified 50 comments per bot. Of Qodo's, 20% were real bugs and 54% were nits, with 12% false positives, and Qodo caused the fewest extra pushes. The team turned Copilot review off and made Qodo the first reviewer, running once on open with only high-severity findings posted inline.
"Copilot is the noise source; Bugbot finds the most real bugs; Qodo converges but its false positives were all process "Rule violations"."
Positive
Ambrish Vadnerkar, Medium
How I Added AI Code Reviews to Our GitLab Pipeline Using PR-AgentPersonal blog post, Medium
Ambrish Vadnerkar found generated descriptions the biggest benefit after a few months of PR-Agent on every merge request. It caught a hardcoded API key, but noise persisted. Vadnerkar worries that the team will rely on it too much.
Mixed
dbudwin, PR-Agent user
Is it possible to remember code suggestions from the previous push?GitHub issue
dbudwin's PR-Agent setup repeated suggestions on every push. Maintainers later added a persistent inline comments setting to skip suggestions already posted.
Mixed or negative, 4
Mixed
cptkoolbeenz, DAQiFi firmware team member
review config: Qodo Rule 232249 contradicts CLAUDE.md's atomicity rule (and its own cited precedent #1028)GitHub issue
A Qodo rule contradicts the project's documented atomicity rule and will fire on every related PR. The team declined one valid lock-free fix because of cost.
Mixed
myakove, mtv-api-tests collaborator
Red Hat QE reports a high-severity Qodo violation on every documentation PR. The rule flags committed generated files that .pr_agent.toml already excludes from review.
Negative
Mark H., reviewer on G2
Underwhelming Experience with Limited Features for the PriceG2 review, syndicated on AWS Marketplace
Mark H. rates Qodo 1 of 5 and prefers Greptile after four months without visible product progress. They cite weak analytics, unintuitive generated rules, missing repository context, and broken GitHub sign-up.
Negative
swishfishdemolish, Indie Hackers member
We Ran 11 AI PR Bots (and kept 1)Indie Hackers post
The Indie Hackers member ranked Qodo eighth among 11 review bots tested on a production repository. It hallucinated often and lacked settings to cut the noise, though it was relatively cheap.
GILD's fact checks, which don't count as sources
Checked Oct 12, 2026.Qodo reads .pr_agent.toml at repository, project, group, or organization level. The highest-precedence setting wins. Qodo recommends portal configuration. Review Standards, the rules and skills it enforces, are in beta. Configuration file
Open them in the recordEvery concern practitioners raise, most raised first.
Counted for Engineering leaders and teams.
Raised as a concern, 8
Positive
Jim Freeman, founder at Punt Labs
ci(review): pilot Qodo-first review with bounded rounds (vox-2q95)GitHub pull request on punt-labs/vox
Across 923 pull requests, Punt Labs hand-classified 50 comments per bot. Of Qodo's, 20% were real bugs and 54% were nits, with 12% false positives, and Qodo caused the fewest extra pushes. The team turned Copilot review off and made Qodo the first reviewer, running once on open with only high-severity findings posted inline.
"Copilot is the noise source; Bugbot finds the most real bugs; Qodo converges but its false positives were all process "Rule violations"."
Mixed
myakove, mtv-api-tests collaborator
Red Hat QE reports a high-severity Qodo violation on every documentation PR. The rule flags committed generated files that .pr_agent.toml already excludes from review.
Mixed
Rahul S., reviewer on G2
Great Cross-Repo PR Insights, Though Severity Flags Can Miss the MarkG2 review, syndicated on AWS Marketplace
Rahul S. rates Qodo 3 of 5. Their team uses it as an extra check on every pull request and values its cross-repository context. It sometimes flags trivial issues at high severity.
Mixed
cptkoolbeenz, DAQiFi firmware team member
review config: Qodo Rule 232249 contradicts CLAUDE.md's atomicity rule (and its own cited precedent #1028)GitHub issue
A Qodo rule contradicts the project's documented atomicity rule and will fire on every related PR. The team declined one valid lock-free fix because of cost.
Negative
swishfishdemolish, Indie Hackers member
We Ran 11 AI PR Bots (and kept 1)Indie Hackers post
The Indie Hackers member ranked Qodo eighth among 11 review bots tested on a production repository. It hallucinated often and lacked settings to cut the noise, though it was relatively cheap.
Mixed
dbudwin, PR-Agent user
Is it possible to remember code suggestions from the previous push?GitHub issue
dbudwin's PR-Agent setup repeated suggestions on every push. Maintainers later added a persistent inline comments setting to skip suggestions already posted.
Mixed
Umut Cihan, Bilkent University, with co-authors from Beko
Automated Code Review In PracticeConference talk page, ICSE 2025 Software Engineering in Practice track
At Beko, developers resolved 73.8% of the comments from a reviewer built on open source Qodo PR-Agent. Average pull request closure time rose from 5 hours 52 minutes to 8 hours 20 minutes. Developers credited faster bug detection and reported faulty reviews and irrelevant comments.
"%73.8 of automated code review comments were labeled as resolved. However, the overall average pull request closure duration increased from five hours 52 minutes to eight hours 20 minutes…"
Positive
Ambrish Vadnerkar, Medium
How I Added AI Code Reviews to Our GitLab Pipeline Using PR-AgentPersonal blog post, Medium
Ambrish Vadnerkar found generated descriptions the biggest benefit after a few months of PR-Agent on every merge request. It caught a hardcoded API key, but noise persisted. Vadnerkar worries that the team will rely on it too much.
Discussed without concern, 2
Mixed
Veli Karakaya, Bilkent University, with Baykal Mehmet Uçar of Beko
Understanding the Limits of Automated Evaluation for Code Review Bots in PracticeResearch paper, arXiv (EASE 2026)
Of 2,604 labeled comments from the reviewer Beko built on open source PR-Agent and ran on GPT-4 Turbo, engineers marked 1,733 (about 66%) fixed and 871 (about 34%) wontFix. Beko's director of software engineering says those labels reflect delivery pressure as well as comment quality, so the bots need continuous monitoring.
"We analyze an industrial dataset from Beko comprising 2,604 bot-generated PR comments, each labeled by software engineers as fixed/wontFix."
Positive
Vivaan K., reviewer on G2
Big-Picture Code Reviews with Qodo, Great for Catching Logic and Dependency IssuesG2 review, syndicated on AWS Marketplace
Vivaan K. rates Qodo 4 of 5. The data engineer values repository-wide checks for logic and dependency issues before merge. Judging findings still requires pipeline and business knowledge.
GILD's fact checks, which don't count as sources
Checked Oct 12, 2026.On GitHub and GitLab, inline_comments_severity_threshold keeps lower-severity findings in the summary. Relevance, in beta, rates each finding based on how the team handled similar findings before. Severity thresholds
Open them in the recordCounted for Engineering leaders and teams.
Raised as a concern, 1
Negative
mattvv, Hacker News commenter
Comment on: A real-world benchmark for AI code reviewHacker News comment
mattvv ruled Qodo out at US$30 per developer a month with 20 pull requests per user. Some developers on their team open that many in a day.
Discussed without concern, 4
Positive
Titus Fortner, Technical Leadership Committee member at Selenium
[build] automatically generate Qodo reviews from PRs by committersGitHub pull request on SeleniumHQ/selenium
Because Qodo pauses reviews for authors without a seat, most Selenium pull requests went unreviewed unless someone with a seat asked. The project now posts the Qodo commands from a CI account that holds a seat, so every committer pull request gets a description and a review on open and on each push.
"Qodo pauses reviews for PR authors without a Qodo seat, so most PRs only get reviewed when someone with a seat comments…"
Mixed
Chris Vancoillie (eleshar), LightSpeed WordPress Development
task: qodo-pr-agent - Validate pilot credential boundary and activationGitHub issue
LightSpeed caps its self-hosted PR-Agent model key at US$20 a month, about 140 automatic runs by its estimate. The release pins an older LiteLLM model map, which prevents a move to Sonnet 5.5.
Positive
Ambrish Vadnerkar, Medium
How I Added AI Code Reviews to Our GitLab Pipeline Using PR-AgentPersonal blog post, Medium
Ambrish Vadnerkar found generated descriptions the biggest benefit after a few months of PR-Agent on every merge request. It caught a hardcoded API key, but noise persisted. Vadnerkar worries that the team will rely on it too much.
Negative
swishfishdemolish, Indie Hackers member
We Ran 11 AI PR Bots (and kept 1)Indie Hackers post
The Indie Hackers member ranked Qodo eighth among 11 review bots tested on a production repository. It hallucinated often and lacked settings to cut the noise, though it was relatively cheap.
GILD's fact checks, which don't count as sources
Checked Oct 12, 2026.Each workspace starts with a 14-day trial. Usage stops at trial end unless paid billing is active. The workspace shares credits, and overage uses the base rate up to a monthly cap. Free open source reviews require a public GitHub repository with 200 or more stars, or an organization that has one. Qodo pricing & usage
Open them in the recordStatements this group's sources agree on, then the ones close to agreement.
Qodo provides a first pass that speeds up human review. People own the final review and merge decision.
5 supporting sourcesQuestions the sources split on, with the count on each side.
Evidence remains split
Kept it, 5 sources
These sources kept Qodo on every pull request for its summaries and bug catches, as the primary reviewer or a fallback.
Dropped it, 3 sources
These sources removed Qodo, most often after the trial ended or reviews stopped. A few found its comments unhelpful.
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 Engineering leaders and teams.
24 sources. Teams and open source projects report how they run Qodo reviewers, what they measure, and whether they keep them.
29 sources. Developers running Qodo on their own repositories report on trial endings, credit pricing, and missing reviews.
53 sources. All accounts together: teams running open source PR-Agent and developers using Qodo's hosted reviewer, whose costs and limits differ.
53 sources from practitioners, by the context they work in. Filled rows count for this group.
Emphasizes Trial endings, pricing, and missing reviews
Emphasizes Summaries, configuration, and noise
Emphasizes Adoption rate and lead time
Emphasizes Cost of self-hosting at scale
What changes with scale. Teams describe noise, setup, and whether the reviewer earns a permanent place. Teams that self-host PR-Agent also report model cost and maintenance. Individual developers describe bug catches and what happens when trials end or reviews stop.
Where this group's sources were published.
Counted for Engineering leaders and teams, from 24 sources. Change group
39 sources for this group, 15 of them documentation
GILD keeps older evidence when it still applies. Accounts tied to renamed products, retired plans, or open source PR-Agent remain in the history as 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 68 sourcesAbout this read
A running review of what engineering teams report about Qodo, 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 3 of 23, individual developers 0 of 26, all practitioners 3 of 49.
Each row opens its sources for the group you are reading.
| Finding | Engineering leaders and teams | Individual developers | All practitioners | Sources |
|---|---|---|---|---|
| Current readCurrent sentiment | 9 positive, 9 mixed, 5 negative | 10 positive, 11 mixed, 5 negative | 19 positive, 20 mixed, 10 negative | Sources |
| PraisePR summaries and effort labels | 3 of 3 positive (under the minimum) | 1 of 1 positive (under the minimum) | 4 of 4 positive (under the minimum) | Sources |
| PraiseReal bug catches | 7 of 7 positive | 11 of 11 positive | 18 of 18 positive | Sources |
| PraiseSuggested fixes | 1 of 2 positive (under the minimum) | 1 of 5 positive | 2 of 7 positive | Sources |
| PraiseConfigurable review rules | 3 of 7 positive | 5 of 5 positive | 8 of 12 positive | Sources |
| PraiseGit platform coverage | 0 of 1 positive (under the minimum) | 0 of 0 positive (under the minimum) | 0 of 1 positive (under the minimum) | Sources |
| PraiseValue for Money | 0 of 1 positive (under the minimum) | 1 of 3 positive (under the minimum) | 1 of 4 positive (under the minimum) | Sources |
| ConcernDashboard and setup friction | 1 of 1 raise it (under the minimum) | 1 of 1 raise it (under the minimum) | 2 of 2 raise it (under the minimum) | Sources |
| ConcernComment noise and wrong findings | 8 of 10 raise it | 6 of 8 raise it | 14 of 18 raise it | Sources |
| ConcernCredits, trials, and price | 1 of 5 raise it | 5 of 6 raise it | 6 of 11 raise it | Sources |
| ConcernMissing reviews and setup | 3 of 3 raise it (under the minimum) | 6 of 6 raise it | 9 of 9 raise it | Sources |
| ConcernSecurity and data exposure | 0 of 2 raise it (under the minimum) | 2 of 2 raise it (under the minimum) | 2 of 4 raise it (under the minimum) | Sources |
| ConsensusA first pass, with humans owning the merge | 5 supporting | 3 supporting (under the minimum) | 8 supporting (below threshold) | Sources |
| ConsensusSelf-hosting on your own model keys matters | 4 supporting (under the minimum) | 2 supporting (under the minimum) | 6 supporting (below threshold) | Sources |
| DisagreementDoes Qodo's reviewer earn a permanent place on the pull request? | Kept it 5, Dropped it 3 | Kept it 3, Dropped it 3 | Kept it 8, Dropped it 6 | Sources |
| DisagreementCompared hands-on, do practitioners prefer Qodo or another reviewer? | Prefer Qodo 1, Prefer another 5 (under the minimum) | Prefer Qodo 1, Prefer another 2 (under the minimum) | Prefer Qodo 2, Prefer another 7 | Sources |
| Current readEvidence mix | 39 sources | 44 sources | 68 sources | Sources |
GILD's checks of claims against official documentation, kept apart from the source counts.
