chore: add verification hierarchy guidance#238
Conversation
Agents should always find a way to verify their work rather than assuming correctness. Adds a prioritised verification hierarchy: tools > browser > primary sources > self-review > ask user.
Summary of ChangesHello @marcusquinn, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request introduces a standardized and prioritized approach to verification for agents. The changes aim to address instances where agents might skip verification or rely on limited methods, by providing a clear hierarchy of steps to confirm their work. This ensures a more robust and consistent quality assurance process across all agent outputs, ultimately leading to more reliable results. Highlights
Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here. You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension. Footnotes
|
|
Caution Review failedThe pull request is closed. WalkthroughAdds verification guidance to two agent configuration files: a five-step "Verification hierarchy" checklist in Changes
Estimated code review effort🎯 1 (Trivial) | ⏱️ ~3 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
🤖 Augment PR SummarySummary: Adds agent-facing guidance to consistently verify work using a prioritized hierarchy. Changes: Updates 🤖 Was this summary useful? React with 👍 or 👎 |
![augmentcode[bot]](https://avatars.githubusercontent.com/in/1027498?s=60&v=4){kind=small}
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
There was a problem hiding this comment.
Code Review
This pull request introduces a verification hierarchy for the agent, adding it to the build-plus.md documentation and as a new principle in the build.txt prompt. The documentation change is clear and well-structured. However, the corresponding prompt added to build.txt is an incomplete and somewhat inconsistent summary of the full hierarchy. I've suggested a more concise, yet complete, version for the prompt to ensure the agent has the full context and follows the intended verification process.
.agent/prompts/build.txt
Outdated
| - Do not add code comments unless asked. Focus on why, not what. | ||
| - After code changes, run lint/typecheck commands if available (check README or configs). | ||
| - When making changes, read surrounding context first to ensure idiomatic integration. | ||
| - Always find a way to verify your work. Prefer in order: run tests/linters, use browser tools to check UI, read primary sources (docs, APIs), or review the output yourself. If none of these give confidence, ask the user how to verify. |
There was a problem hiding this comment.
This summary of the verification hierarchy is inconsistent with the full version documented in .agent/build-plus.md and the PR description. It omits key verification tools like 'type checkers' and 'build commands', as well as 'git log' as a primary source. Additionally, the mention of 'tests/linters' is slightly redundant and inconsistent with line 106 which specifies 'lint/typecheck'. To ensure the agent has clear and complete instructions, consider a more structured and comprehensive summary that aligns with the documented hierarchy.
- Always verify your work. Hierarchy: 1. Tools (tests, lint, typecheck, build). 2. Browser UI. 3. Docs/APIs/git-log. 4. Self-review. 5. Ask user.
🔍 Code Quality Report�[0;35m[MONITOR]�[0m Code Review Monitoring Report �[0;34m[INFO]�[0m Latest Quality Status: �[0;34m[INFO]�[0m Recent monitoring activity: 📈 Current Quality Metrics
Generated on: Tue Jan 27 01:59:38 UTC 2026 Generated by AI DevOps Framework Code Review Monitoring |
Address Gemini review feedback: include type checkers, build commands, and git log to match the documented hierarchy in build-plus.md.
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
|
🔍 Code Quality Report�[0;35m[MONITOR]�[0m Code Review Monitoring Report �[0;34m[INFO]�[0m Latest Quality Status: �[0;34m[INFO]�[0m Recent monitoring activity: 📈 Current Quality Metrics
Generated on: Tue Jan 27 02:02:47 UTC 2026 Generated by AI DevOps Framework Code Review Monitoring |
1 participant
Summary
build.txt(applies to all agents)Verification Hierarchy
git log)Motivation
Agents sometimes skip verification entirely or only verify through tests. This ensures they always find some way to confirm their work is correct, escalating to the user when automated options are uncertain.
Summary by CodeRabbit
✏️ Tip: You can customize this high-level summary in your review settings.