- Home
- Skills
- Code Quality & Review
- AI Generated Code Verification Engineer Skills Pro
Works with the AI Agents you already use
AI Generated Code Verification Engineer Skills Pro
Independent verification engineer for auditing AI-generated code through evidence-based tests and security checks.
$19.90
AI Generated Code Verification Engineer Skills Pro
Example session with this skill installed
Verify the recent changes in this PR. I've added a new JWT auth middleware. Run the tests, check for security risks, and give me a full verification report. I authorize read-only checks.
- Read your context and instructions
- Compiled the ai generated code
I have completed the verification of the JWT authentication middleware. I inspected the diff in src/middleware/auth.ts, ran the Vitest suite, and performed a manual static analysis of the token validation logic.
AI-Generated Code Verification Report
Executive verdict
- Verdict: VERIFIED WITH RISKS
- Scope: JWT Middleware implementation and associated unit tests
- Overall confidence: Medium
- Merge recommendation: Approve with conditions (see findings)
Requirements traceability
| Requirement | Evidence | Status |
|---|---|---|
| Valid tokens grant access | npm run test passed 4/4 cases in auth.test.ts | PASS |
| Expired tokens are rejected | Verified in auth.test.ts line 42 | PASS |
| Secret is not hardcoded | src/config.ts uses process.env.JWT_SECRET | PASS |
| Handle missing header | Test case should return 401 if no header | PASS |
Checks executed
npm run test src/middleware/auth.test.ts: Passed (Exit 0)npx eslint src/middleware/auth.ts: No errors (Exit 0)grep -r "JWT_SECRET" .: Confirmed environment variable usage
Findings
- Severity: Medium
- Title: Missing Algorithm Enforcement
- File:
src/middleware/auth.ts:18 - Evidence:
jwt.verify(token, secret)is called without specifyingalgorithms: ['HS256']. - Impact: Potential algorithm confusion attacks if the library defaults are permissive.
- Remediation: Explicitly pass the allowed algorithm in the verify options.
Gaps and limitations
Integration tests with the live Redis blacklist were not run as the environment was unavailable. Verification is limited to unit-level logic.
Next steps
- Update
jwt.verifyto explicitly includealgorithms: ['HS256']. - Provide a mock for the Redis service to allow full integration verification.
- Ensure
JWT_SECRETis set in the CI environment secrets.
Connects securely to your tools. The creator never sees your data.
About this skill
AI-generated code often looks correct but fails silently in production or introduces subtle security risks. This skill acts as an independent verification engineer that audits code changes, validates requirements against implementation, and runs evidence-based tests to confirm behavior. It moves beyond simple "does it compile" checks to provide a defensible merge recommendation.
What it does
- Requirement traceability maps implementation details directly to acceptance criteria to ensure nothing was missed.
- Evidence-based auditing verifies claims by running actual commands and inspecting exit statuses rather than assuming success.
- Security-focused review scans for injection risks, broken access control, and unsafe secret handling.
- Test quality evaluation identifies weak assertions, mocked-out behavior, and deterministic failures that hide bugs.
- Verification reporting generates structured reports with verdicts ranging from Verified to Rejected based on objective data.
How it works
- Scope and baseline inspection identifies the specific diff, repository constraints, and project-specific test commands.
- Matrix creation converts requirements into a requirement-to-evidence matrix with defined pass/fail states.
- Check execution runs unit, integration, and static analysis tools discovered from the environment.
- Final reporting delivers an AI-Generated Code Verification Report including an executive verdict and prioritized findings.
Frameworks & tools
Works with any language or framework including JavaScript, Python, Go, Rust, and Java. It discovers project-specific commands from package.json, Makefile, README, and CI configuration files.
Why this beats prompting it yourself
Standard prompts often suffer from sycophancy, where the AI confirms its own logic or ignores edge cases to appear helpful. This skill enforces a skeptical "evidence over confidence" protocol that requires actual command output and identifies "not verified" states rather than hallucinating success.
Use cases
- AI agent supervision to audit the output of coding agents before merging their PRs.
- CI/CD gatekeeping to determine if a change has enough test coverage and evidence to pass a release gate.
- Security sanity checks for identifying common vulnerabilities in third-party or AI-generated patches.
- Bug fix validation to ensure a regression test actually fails before the fix and passes after.
Known limitations
Does not perform live system exploitation or intrusive network testing. Requires clear acceptance criteria and access to a local environment where tests can be safely executed.
Requires access to local test runners and clear acceptance criteria. Will not execute destructive operations or network calls without explicit authorization. Verdicts are limited to local evidence.
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
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 16 days ago