An advanced AI code review agent that performs regression-focused, context-aware code analysis for local diffs, commits, GitHub PRs, and GitLab MRs to streamline merge decisions.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install code-review-promax
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install code-review-promax using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is code-review-ProMax?
code-review-ProMax is a senior code review agent designed to perform contextual, regression-risk-oriented code reviews. Unlike simple linters, this agent acts as a seasoned software engineer evaluating changes on correctness, stability, security, upstream/downstream impacts, and architectural compatibility.
By integrating seamlessly into developer workflows, it is one of the most powerful utilities available in Openclaw Skills. It handles multiple input sources, including raw diffs, commit hashes, local Git changes, GitHub Pull Requests, and GitLab Merge Requests. The agent automatically infers the modification intent (such as a bug fix, refactoring, or performance optimization) and adjusts its evaluation depth and criteria accordingly.
code-review-ProMax Use Cases
- Automating local code quality checks before pushing changes to remote branches.
- Reviewing incoming GitHub Pull Requests or GitLab Merge Requests directly from the terminal or AI environment.
- Performing targeted regression risk assessments for large, complex code refactoring tasks.
- Generating structured, copy-pasteable repair instructions to quickly address code quality issues via specialized sub-skills.
How code-review-ProMax Works
- Acquire Latest Changes: The agent checks the workspace for local changes using git commands or fetches pull request/merge request diffs via GitHub/GitLab APIs.
- Infer Change Intent: It analyzes the context and categorizes the intent (such as Bug Fix, Refactor, Feature, or Hotfix) to calibrate the review rules.
- Perform Multi-Dimensional Review: It evaluates code along 10 dimensions, including correctness, boundary limits, concurrency, API compatibility, and security compliance.
- Build Risk Deduction Chains: For critical or high-severity issues, the agent maps out the regression path (Change -> Direct Impact -> Cascading Impact -> Worst-Case Scenario).
- Deliver Verdict & Fix Tasks: It computes a definitive verdict (such as Direct Merge, Merge after Fix, or Verify Further) and generates structured, copy-pasteable fix instructions.
code-review-ProMax Setup
To deploy code-review-ProMax within your Openclaw Skills environment, configure your workspace as follows:
- Ensure you have the GitHub CLI ('gh') installed or configure a 'GITHUB_TOKEN' for repository-level API access:
export GITHUB_TOKEN='your-github-token'
gh auth login
- For GitLab integrations, define the appropriate environment variables to allow API calls to private instances:
export GITLAB_PRIVATE_TOKEN='your-gitlab-token'
- Execute the skill triggers (such as 'review this PR' or 'CR local changes') inside your active terminal to initialize the review process.
code-review-ProMax Data Schema & Taxonomy
The skill leverages specialized sub-skills to handle different execution modes:
| Directory / File | Description | Purpose |
|---|---|---|
fix/SKILL.md |
Direct Fix Sub-Skill | Handles automatic code corrections when the user requests direct fixes. |
focused/SKILL.md |
Focused Review Sub-Skill | Performs deep analyses on specific code paths or designated interest areas. |
iterative/SKILL.md |
Iterative Review Sub-Skill | Tracks changes across multiple revision rounds, detecting suppression patterns. |
Report Output Schema
Reviews are returned as a structured Markdown document containing:
- Metadata: Risk Rating (Low/Medium/High), Confidence Level (Sure/Possible/Suspected), Verdict.
- Categorized Sections:
- No-Impact Changes (pure formatting, copy edits)
- Advisory Issues (non-blocking suggestions)
- Blockers / Must-Fix Issues (with complete cascading risk paths)
- Task Block: Code blocks tagged as
## Code Review Repair Taskto facilitate automatic ingestion by other agents.
code-review-ProMax Advanced Features
- Intent-Driven Evaluation: Dynamically shifts assessment metrics depending on whether changes represent a hotfix, refactor, or security patch.
- Deep Context Inspection: Moves beyond superficial diff analysis by pulling surrounding function chains, module declarations, and dependencies.
- Direct Repair Action Mode: Seamlessly switches to "Direct Fix Mode" using
fix/SKILL.mdto automatically apply corrections to identified issues. - Strict SEO & Web Compliance Review: Runs dedicated audits on web-facing files to flag issues like cloaking, dynamic JS injects, and keyword stuffing.
SKILL.md
Loading
Related Openclaw Skills
Vylai Instance Manager
An automated command-line toolset to create, query, scale, and clean up high-performance GPU instances on the Vylai cloud platform.
Upstream Material Analysis Framework
A systematic research framework designed to analyze non-consumer-facing upstream material providers within complex supply chains.
Daily Intel Report
An automated market intelligence and bidding aggregator that delivers structured daily reports on AI industry updates and government tenders directly to Feishu or email.
legal-verify
An advanced verification tool that extracts legal citations from draft texts and validates them against authoritative databases to eliminate LLM hallucinations.
Nexus Wage Widget
A highly customizable, responsive glassmorphism HTML widget combining real-time earnings tracking with a Pomodoro focus timer.
Bocha Web Search A2M
A professional AI-optimized web search skill implementing HTTP 402 A2M protocol to deliver accurate, complete search results and summaries.
Featured*
