- Home
- Skills
- Agents & Orchestration
- safe ai coding workflow guardian
safe ai coding workflow guardian
Enforce verifiable, stage-gated engineering workflows and safety gates for AI-driven codebase modifications.
$5
Works with the AI tools you already use
safe ai coding workflow guardian
Example session with this skill installed
Implement a new JWT-based authentication middleware for the /api/v2 routes and ensure it fails closed on expired tokens.
- Read your context and instructions
- Compiled the safe ai coding
I have completed the safe coding workflow for the requested JWT authentication middleware. All safety gates passed, and implementation was verified stage-by-stage.
Closure Evidence: auth-middleware-jwt-001
Verdict: READY_FOR_HUMAN_REVIEW
Scope Summary
Implemented JwtAuthMiddleware in src/middleware/auth.ts, applied to all /api/v2 routes. The system now enforces RSA256 signature verification and explicitly fails closed on expired or malformed tokens.
Stage Status
| Stage | Status | Notes |
|---|---|---|
| 1 Context | PASS | Mapped routes/api_v2.ts and existing AuthService. |
| 2 Scope | PASS | Defined 4 criteria including expiration handling. |
| 3 Safety | PASS | Verified AUTH_PRIVATE_KEY env var is present. |
| 4 Implementation | PASS | Verified via npm run build. |
| 5 Tests | PASS | 12 unit tests, 4 integration tests green. |
| 6 Quality | PASS | ESLint and Semgrep (security) passed. |
| 7 Risk | PASS | No new attack surface identified. |
Acceptance Criteria Traceability
| Criterion | Test Covering | Status |
|---|---|---|
| Reject expired tokens | tests/auth.spec.ts:L45 | PASS |
| Fail closed on missing header | tests/auth.spec.ts:L60 | PASS |
| Support RSA256 sigs | tests/auth.spec.ts:L82 | PASS |
Open Items
- Performance benchmark under 5k concurrent reqs (owner: DevOps, deadline: 2024-12-15)
Deferred Items (with reason)
- Redis blacklist check (reason: Out of scope for initial JWT implementation; to be added in Phase 2)
Known Limitations
- Does not currently handle token refreshing logic; strictly handles validation.
Next steps
- Review the generated
src/middleware/auth.tslogic. - Execute
git merge feature/jwt-authif the evidence meets requirements. - Update the production environment variables with the new public key.
Connects securely to your tools. The creator never sees your data.
What you get
About this skill
The problem
AI agents often suffer from "hallucinated progress," skipping vital verification steps or introducing scope creep that breaks production environments. Developers lose hours debugging unverified changes that lacked proper safety gates or traceability.
What it does
- Enforces a strict, stage-gated engineering workflow that prevents code implementation until scope and safety plans are locked.
- Generates a context map and risk scan to identify breaking changes before a single line of code is modified.
- Captures raw terminal output for every verification step, ensuring implementation logs are backed by actual tool data.
- Produces a standardized closure evidence artifact that summarizes test coverage, quality checks, and risk assessments for human review.
Frameworks & tools
Works with any CLI-based toolchain including git, npm/jest, pytest, semgrep, and gitleaks. It maps logical verification steps to terminal and file_system operations.
Why this beats prompting it yourself
Standard prompts often fail-open, assuming success when a tool outputs an error. This skill is hard-coded to fail-closed, stopping the entire workflow if a safety gate or test fails, ensuring no unverified code ever reaches the human review stage.
Use cases
- Implementing complex logic changes in legacy systems where regression risk is high.
- Standardizing PR evidence for teams requiring high auditability and compliance.
- Fixing critical bugs that require pre-flight environment checks and database backups.
Known limitations
Requires an environment with terminal access to run verification commands. Not intended for trivial one-line typo fixes or documentation edits.
How to install
Works the same in every agent - Claude, Cursor, Codex, Copilot and 20+ more.
- 1
Download the ZIP
Free skills download straight away. Paid skills unlock right after purchase.
- 2
Unzip into your skills folder
Every agent reads skills from one folder on your machine. Drop the unzipped folder in there.
- 3
Ask your agent to use it
Restart the agent if it was already running. It picks the skill up automatically - no config needed.
Skills folder by agent
Click the path to copy it. Create the folder if it does not exist yet.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill passes our trust checks below.
Security scanned
Passed our 8-point scan before listing
Fresh listing
Recently published to Agensi
30-day refund
Not a fit? Get your money back
Trust & safety
Security scanned
Verified clean 2 days ago
- Passed all security checks, Safe to install