- Home
- Skills
- Code Quality & Review
- Strict Code Review Pro
Works with the AI Agents you already use
Strict Code Review Pro
Strict, evidence-based code auditor that detects spec drift and enforces verification gates with veto power.
$12
Strict Code Review Pro
Example session with this skill installed
Strict review the current changes against specs/export-users/. Run the project checks.
- Read your context and instructions
- Compiled the strict code review
I read the specs, the diff and the changed code, then ran the project checks. The review is
REJECTED: one test fails and the code disagrees with the spec on the row limit. I wrote review.md and did not edit any code or spec.
Verification commands
| Command | Result |
|---|---|
npm run typecheck | OK |
npm run lint | OK |
npm run build | OK |
npm test | FAIL - 1 failed: serialize > escapes double quotes (expected "a""b", got "a"b") |
Findings
| ID | Severity | Category | File:line | Description | Suggested action |
|---|---|---|---|---|---|
| R1 | high | correctness | src/shared/csv/serialize.ts:18 | Quotes are not doubled, so a value with " yields malformed CSV | Replace " with "" before wrapping the field |
| R2 | high | drift | src/features/exportUsers/api.ts:34 | Limit is 100000 but spec requirement 3 says 50,000 | Fix code to 50,000, or amend the spec if 100,000 is the new intent |
| R3 | medium | security | src/features/exportUsers/api.ts:12 | Handler checks login but not the admin role | Add the admin check and a 403 test (TC6) |
| R4 | low | style | src/entities/user/model.ts:8 | Needless let | Use const |
Next steps
- Fix R1 and R2 (the two blockers) and add the 403 test for R3.
- Re-run the review. It will approve when there are no high findings and the checks are green.
VETO: tests failing and drift at src/features/exportUsers/api.ts:34
Connects securely to your tools. The creator never sees your data.
About this skill
Reviewing code is often too passive, leading to architecture violations and spec drift. This skill acts as a strict auditor with veto power, ensuring every change aligns with your documentation and passes all verification gates before reaching the codebase.
What it does
- Veto power over any implementation that violates architecture or fails project tests.
- Drift detection to identify discrepancies where the code and specifications disagree.
- Evidence-based auditing requiring specific file and line citations for every finding.
- Verification execution by running your project's native typechecks, lints, and test suites.
- Standardized reporting via a generated
review.mdfile with clear severity classifications.
How it works
- Context ingestion reads your specs, designs, tasks, and changed code to establish the ground truth.
- Systematic audit follows your internal review checklist and runs all local verification commands.
- Finding classification labels issues by severity and category, such as security, drift, or correctness.
- Verdict generation applies strict logic to issue an
APPROVEDorREJECTEDstatus based on finding data.
Frameworks & tools
This skill works with any stack by detecting project manifests (package.json, pyproject.toml, etc.) to run standard build and test tools. It uses Git commands to identify local diffs.
Why this beats prompting it yourself
Generic prompts often try to "fix" the code or ignore the underlying specifications. This skill enforces a read-only auditing mindset that prevents the AI from changing the spec to hide a mistake, ensuring 100% alignment with your spec.md.
Use cases
- Pre-merge auditing to catch architectural drift before a pull request is opened.
- Strict compliance checking for security-sensitive or high-stakes implementations.
- Automated verification of task completion against specific design documents.
Known limitations
This skill is strictly read-only for source files and documentation. It cannot fix the issues it finds and requires a spec.md or similar documentation to detect drift.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Code Quality & Review.
Read-only on source and specs. Cannot fix found issues. Requires documentation (spec.md/design.md) for drift detection. Verification requires detectable or provided project commands.
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 buying a skill.
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 19 days ago
- Software
Git