Re: [PR] feat: Add `tvm-ffi-code-review` Claude Skill [tvm-ffi]
junrushao closed pull request #450: feat: Add `tvm-ffi-code-review` Claude Skill URL: https://github.com/apache/tvm-ffi/pull/450 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] - To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
Re: [PR] feat: Add `tvm-ffi-code-review` Claude Skill [tvm-ffi]
junrushao commented on PR #450: URL: https://github.com/apache/tvm-ffi/pull/450#issuecomment-3906062835 Closing as it's not specific to TVM-FFI. Would be nice to be a global skill -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] - To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
Re: [PR] feat: Add `tvm-ffi-code-review` Claude Skill [tvm-ffi]
junrushao commented on code in PR #450: URL: https://github.com/apache/tvm-ffi/pull/450#discussion_r2808232443 ## .claude/skills/tvm-ffi-code-review/SKILL.md: ## @@ -0,0 +1,239 @@ + + + + + + + + + + + + + + + + + +--- +name: tvm-ffi-code-review +description: Run parallel code reviews using Claude Code and OpenAI Codex reviewers. Produces a unified, prioritized review report with actionable findings from multiple AI models. +disable-model-invocation: true +argument-hint: "[pr | branch: | commit: | staged | unstaged]" +allowed-tools: Bash(git *), Bash(gh *), Read, Grep, Glob, Task, AskUserQuestion +--- + +# Multi-Model Code Review + +Review code changes using two independent AI reviewers in parallel — **Claude Code** and **OpenAI Codex** — then synthesize their findings into a single prioritized report. + +## Prerequisites + +- **Codex CLI** must be installed and authenticated (`npm install -g @openai/codex` or equivalent). If unavailable, the skill gracefully falls back to Claude-only review. + +## Step 1: Determine review scope + +Parse `$ARGUMENTS` to determine what to review. If the argument is empty or ambiguous, prompt the user with `AskUserQuestion` to choose a scope. + +### Supported scopes + +| Argument | Diff command | Description | +|--|-|-| +| `pr` (default) | `git diff $(git merge-base HEAD )...HEAD` | All changes in the current PR/branch since it diverged from the main branch. Auto-detects the main branch (`main` or `master`). | +| `branch:` | `git diff ...HEAD` | Changes relative to the named branch. | +| `commit:` | `git diff ...HEAD` | Changes since the given commit. | +| `staged` | `git diff --cached` | Only staged (indexed) changes. | +| `unstaged` | `git diff` | Only unstaged working-tree changes. | + +### Scope resolution logic + +1. If `$ARGUMENTS` is empty, default to `pr` scope. +2. If `$ARGUMENTS` matches one of the keywords above, use that scope. +3. If `$ARGUMENTS` looks like a branch name or commit SHA (doesn't match a keyword), treat it as `branch:` for backward compatibility. +4. If the argument is ambiguous or the ref doesn't exist, ask the user: + +``` +AskUserQuestion: + question: "What would you like to review?" + options: +- "Current PR (all commits since diverging from main)" +- "Against a specific branch" +- "Since a specific commit" +- "Staged changes only" +``` + +### Gather the diff + +Once the scope is resolved, run: + +```bash +git diff --stat +git diff --unified=5 +``` + +Store the diff output and the list of changed files. If the diff exceeds ~5000 lines, identify the most critical files and note that the review covers a subset. + +## Step 2: Launch both reviewers in parallel + +Use the **Task** tool to launch **both** reviewers simultaneously — two parallel Task calls in a single response. + +### Claude Code Reviewer + +``` +Task(subagent_type: "claude-code-reviewer", model: "opus") +``` + +- **Model**: Claude Opus 4.6 (`claude-opus-4-6`) with high reasoning effort. +- Prompt the subagent with: + - The full unified diff + - The list of changed files + - The shared review instruction below + +### Codex Code Reviewer + +``` +Task(subagent_type: "codex-code-reviewer") +``` + +- **Model**: GPT-5.3 Codex (`gpt-5.3-codex-xhigh`). Pass this as a model hint in the prompt to the subagent: "Use model gpt-5.3-codex-xhigh for this review." Review Comment: again. someone please fix gemini -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] - To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
Re: [PR] feat: Add `tvm-ffi-code-review` Claude Skill [tvm-ffi]
junrushao commented on code in PR #450: URL: https://github.com/apache/tvm-ffi/pull/450#discussion_r2808231788 ## .claude/skills/tvm-ffi-code-review/SKILL.md: ## @@ -0,0 +1,239 @@ + + + + + + + + + + + + + + + + + +--- +name: tvm-ffi-code-review +description: Run parallel code reviews using Claude Code and OpenAI Codex reviewers. Produces a unified, prioritized review report with actionable findings from multiple AI models. +disable-model-invocation: true +argument-hint: "[pr | branch: | commit: | staged | unstaged]" +allowed-tools: Bash(git *), Bash(gh *), Read, Grep, Glob, Task, AskUserQuestion +--- + +# Multi-Model Code Review + +Review code changes using two independent AI reviewers in parallel — **Claude Code** and **OpenAI Codex** — then synthesize their findings into a single prioritized report. + +## Prerequisites + +- **Codex CLI** must be installed and authenticated (`npm install -g @openai/codex` or equivalent). If unavailable, the skill gracefully falls back to Claude-only review. Review Comment: worst hallucination i've ever seen :/ -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] - To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
Re: [PR] feat: Add `tvm-ffi-code-review` Claude Skill [tvm-ffi]
gemini-code-assist[bot] commented on code in PR #450: URL: https://github.com/apache/tvm-ffi/pull/450#discussion_r2808229141 ## .claude/skills/tvm-ffi-code-review/SKILL.md: ## @@ -0,0 +1,239 @@ + + + + + + + + + + + + + + + + + +--- +name: tvm-ffi-code-review +description: Run parallel code reviews using Claude Code and OpenAI Codex reviewers. Produces a unified, prioritized review report with actionable findings from multiple AI models. +disable-model-invocation: true +argument-hint: "[pr | branch: | commit: | staged | unstaged]" +allowed-tools: Bash(git *), Bash(gh *), Read, Grep, Glob, Task, AskUserQuestion +--- + +# Multi-Model Code Review + +Review code changes using two independent AI reviewers in parallel — **Claude Code** and **OpenAI Codex** — then synthesize their findings into a single prioritized report. + +## Prerequisites + +- **Codex CLI** must be installed and authenticated (`npm install -g @openai/codex` or equivalent). If unavailable, the skill gracefully falls back to Claude-only review. Review Comment:  The prerequisite for the Codex CLI appears to be incorrect. The npm package `@openai/codex` is not an official or widely-used tool, and the original Codex models have been superseded. This instruction is likely to fail for users and should be updated to specify a verifiable tool and a current, available OpenAI model (e.g., a GPT-4 variant) and the correct method to invoke it. ## .claude/skills/tvm-ffi-code-review/SKILL.md: ## @@ -0,0 +1,239 @@ + + + + + + + + + + + + + + + + + +--- +name: tvm-ffi-code-review +description: Run parallel code reviews using Claude Code and OpenAI Codex reviewers. Produces a unified, prioritized review report with actionable findings from multiple AI models. +disable-model-invocation: true +argument-hint: "[pr | branch: | commit: | staged | unstaged]" +allowed-tools: Bash(git *), Bash(gh *), Read, Grep, Glob, Task, AskUserQuestion +--- + +# Multi-Model Code Review + +Review code changes using two independent AI reviewers in parallel — **Claude Code** and **OpenAI Codex** — then synthesize their findings into a single prioritized report. + +## Prerequisites + +- **Codex CLI** must be installed and authenticated (`npm install -g @openai/codex` or equivalent). If unavailable, the skill gracefully falls back to Claude-only review. + +## Step 1: Determine review scope + +Parse `$ARGUMENTS` to determine what to review. If the argument is empty or ambiguous, prompt the user with `AskUserQuestion` to choose a scope. + +### Supported scopes + +| Argument | Diff command | Description | +|--|-|-| +| `pr` (default) | `git diff $(git merge-base HEAD )...HEAD` | All changes in the current PR/branch since it diverged from the main branch. Auto-detects the main branch (`main` or `master`). | +| `branch:` | `git diff ...HEAD` | Changes relative to the named branch. | +| `commit:` | `git diff ...HEAD` | Changes since the given commit. | +| `staged` | `git diff --cached` | Only staged (indexed) changes. | +| `unstaged` | `git diff` | Only unstaged working-tree changes. | + +### Scope resolution logic + +1. If `$ARGUMENTS` is empty, default to `pr` scope. +2. If `$ARGUMENTS` matches one of the keywords above, use that scope. +3. If `$ARGUMENTS` looks like a branch name or commit SHA (doesn't match a keyword), treat it as `branch:` for backward compatibility. +4. If the argument is ambiguous or the ref doesn't exist, ask the user: + +``` +AskUserQuestion: + question: "What would you like to review?" + options: +- "Current PR (all commits since diverging from main)" +- "Against a specific branch" +- "Since a specific commit" +- "Staged changes only" +``` + +### Gather the diff + +Once the scope is resolved, run: + +```bash +git diff --stat +git diff --unified=5 +``` + +Store the diff output and the list of changed files. If the diff exceeds ~5000 lines, identify the most critical files and note that the review covers a subset. + +## Step 2: Launch both reviewers in parallel + +Use the **Task** tool to launch **both** reviewers simultaneously — two parallel Task calls in a single response. + +### Claude Code Reviewer + +``` +Task(subagent_type: "claude-code-reviewer", model: "opus") +``` + +- **Model**: Claude Opus 4.6 (`claude-opus-4-6`) with high reasoning effort. +- Prompt the subagent with: + - The full unified diff + - The list of changed files + - The shared review instruction below + +### Codex Code Reviewer + +``` +Task(subagent_type: "codex-code-reviewer") +``` + +- **Model**: GPT-5.3 Codex (`gpt-5.3-codex-xhigh`). Pass this as a model hint in the prompt to the subagent: "Use model gpt-5.3-codex-xhigh for this review." +- Prompt the subagent with: + - The full unified diff + - The list of changed files + - The shared review instruction below + +### Shared review instruction + +
Re: [PR] feat: Add `tvm-ffi-code-review` Claude Skill [tvm-ffi]
junrushao commented on PR #450: URL: https://github.com/apache/tvm-ffi/pull/450#issuecomment-3902954653 Example: https://github.com/apache/tvm-ffi/pull/449#pullrequestreview-3802989061 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] - To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
Re: [PR] feat: Add `tvm-ffi-code-review` Claude Skill [tvm-ffi]
gemini-code-assist[bot] commented on PR #450: URL: https://github.com/apache/tvm-ffi/pull/450#issuecomment-3902954595 ## Summary of Changes Hello @junrushao, I'm Gemini Code Assist[^1]! 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 new `tvm-ffi-code-review` skill designed to enhance code review processes by leveraging multiple AI models. The skill automates comprehensive code analysis, offering a consolidated and prioritized report of potential issues, and streamlines feedback delivery through optional GitHub integration. ### Highlights * **Parallel AI Review**: Introduced a new skill that performs parallel code reviews using both Claude Code (Opus 4.6) and OpenAI Codex (GPT-5.3) models. * **Unified Report Generation**: Synthesizes findings from both AI models into a single, prioritized report, detailing consensus findings and divergences. * **Flexible Review Scope**: Supports various review scopes including entire PRs, specific branches, commits, staged changes, or unstaged changes, with user prompting for ambiguous cases. * **Comprehensive Review Checklist**: Both AI reviewers utilize a shared checklist covering correctness, security, performance, API design, maintainability, concurrency, error handling, and best practices. * **GitHub Integration**: Provides an option to publish the synthesized review report as inline comments directly on GitHub Pull Requests using the `gh api`. 🧠 **New Feature in Public Preview:** You can now enable **Memory** to help **Gemini Code Assist** learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. **Click [here](https://codeassist.google/code-review/login) to enable Memory in your admin console.** Changelog * **.claude/skills/tvm-ffi-code-review/SKILL.md** * Added the definition and detailed instructions for the new `tvm-ffi-code-review` skill. Activity * No specific activity (comments, reviews, or progress updates) has been recorded for this pull request yet. Using Gemini Code Assist The full guide for Gemini Code Assist can be found on our [documentation page](https://developers.google.com/gemini-code-assist/docs/review-github-code), here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either `/gemini ` or `@gemini-code-assist `. Below is a summary of the supported commands on the current page. Feature | Command | Description --- | --- | --- Code Review | `/gemini review` | Performs a code review for the current pull request in its current state. Pull Request Summary | `/gemini summary` | Provides a summary of the current pull request in its current state. Comment | @gemini-code-assist | Responds in comments when explicitly tagged, both in pull request comments and review comments. Help | `/gemini help` | Displays a list of available commands. 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 `.gemini/` folder in the base of the repository. Detailed instructions can be found [here](https://developers.google.com/gemini-code-assist/docs/customize-gemini-behavior-github). 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 :thumbsup: and :thumbsdown: 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](https://google.qualtrics.com/jfe/form/SV_2cyuGuTWsEw84yG). 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](https://cloud.google.com/products/gemini/code-assist). [^1]: Review the [Privacy Notices](https://policies.google.com/privacy), [Generative AI Prohibited Use Policy](https://policies.google.com/terms/generative-ai/use-policy), [Terms of Service](https://policies.google.com/terms), and learn how to configure Gemini Code Assist in GitHub [here](https://developers.google.com/gemini-code-assist/docs/customize-gemini-behavior-github). Gemini can make mistakes, so double check it and [use code with caution](https://support.google.com/legal/answer/13505487). -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific co
