- Home
- Skills
- Code Quality & Review
- diff sentinel
diff sentinel
Automated git diff auditor for secret detection, breaking API changes, and context-optimized patch packaging.
Free
Works with the AI Agents you already use
diff sentinel
Example session with this skill installed
Analyze this diff for secrets and breaking changes: git diff origin/main...HEAD
- Read your context and instructions
- Compiled the diff sentinel
I have analyzed the provided diff for security leaks, breaking API changes, and code complexity.
Diff Audit Summary
| Category | Status | Finding | Recommendation |
|---|---|---|---|
| Security | [BLOCKED] | SEC-001: Potential AWS Access Key found in src/config/aws.ts:12 | Remove secret and use environment variables or a secret manager. |
| Breaking Changes | [REVIEW] | API-004: New required parameter timeout added to exported function fetchData in lib/network.ts | Make the parameter optional or provide a default value to prevent breaking existing callers. |
| Complexity | [PASS] | Max cyclomatic complexity is 4 in src/utils/parser.ts | No action required. |
| Dependencies | [PASS] | No changes to manifest files detected. | No action required. |
Complexity Breakdown
src/utils/parser.ts: 4 (Low)src/config/aws.ts: 1 (Low)lib/network.ts: 2 (Low)
Status: [BLOCKED]
Critical security finding detected. Merge is blocked until credentials are removed.
Next steps
- Scrub the hardcoded AWS key from
src/config/aws.tsand rotate the credential. - Update the
fetchDatasignature inlib/network.tstotimeout: number = 5000to maintain backward compatibility. - Re-run
analyze_diffto confirm all blocks are cleared.
Connects securely to your tools. The creator never sees your data.
About this skill
Reviewing pull requests or local changes manually often leads to missed secrets, breaking API changes, or bloated LLM prompts. This skill automates the audit of git diffs to detect security risks and structural regressions before they reach your main branch.
What it does
- Secret detection identifies leaked API keys, tokens, and database passwords in the patch.
- Breaking change analysis checks if exported function signatures or interfaces in TS, Python, Go, or Rust have changed in ways that break callers.
- Context optimization strips lockfiles and generated assets from diffs to minimize token usage when sending code to LLMs.
- Dependency auditing flags risks or downgrades in
package.json,requirements.txt, orCargo.toml. - Complexity scoring calculates cyclomatic complexity to pinpoint deeply nested or unmaintainable code blocks.
How it works
- Input the diff by piping
git diffoutput to the tool or passing the patch to theanalyze_diffMCP tool. - Select an audit mode such as a full analysis, a security-only scan, or a dependency check.
- Review the risk status categorized as
[PASS],[REVIEW], or[BLOCKED]based on the findings. - Export findings as a formatted Markdown table for PR comments or a SARIF file for GitHub Code Scanning.
Frameworks & tools
Built for Node.js 22+ environments. It supports git-based workflows and integrates with TypeScript, JavaScript, Python, Go, and Rust for structural analysis.
Why this beats prompting it yourself
Generic LLM prompts often hallucinate security risks or miss subtle breaking changes in function signatures. This skill uses deterministic parsing to calculate complexity and strip noise, ensuring your context window is spent on logic rather than lockfiles.
Use cases
- Scan uncommitted local changes for hardcoded credentials before a push.
- Generate a summary table of code risks to post as a pull request comment.
- Reduce token costs by cleaning large diffs before asking an AI to refactor code.
- Audit package manifest changes for unexpected dependency downgrades or high-risk additions.
Known limitations
Requires Node.js 22 or higher and a local git installation. Structural breaking change analysis is currently limited to TypeScript, JavaScript, Python, Go, and Rust.
Requires Node.js 22+ and local git. Structural breaking change detection is limited to TS/JS, Python, Go, and Rust. Does not support non-git version control systems.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 10 days ago
- Free with an Agensi account
Requires