- Home
- Skills
- Code Quality & Review
- Ship Gate
Works with the AI Agents you already use
Ship Gate
Final go/no-go before merge or release: runs an eight-step gate sequence and returns exactly PASS, PASS_WITH_WARNINGS or BLOCKED, marking anything unproven as UNVERIFIED. Read-only; a pass never authorizes a deploy.
$25
Ship Gate
Example session with this skill installed
Is the JWT auth migration ready to merge?
- Read your context and instructions
- Compiled the ship gate
Illustrative example, synthetic data. File names, findings and outputs below are made up to show the shape of a real run. In actual use the skill reports only evidence supplied or verified in your environment and marks anything unavailable as UNVERIFIED.
BASELINE: branch auth-jwt @ a1b2c3 vs main; scope = all 4 auth endpoints + middleware.
VERIFIED: lint/type/build PASS; auth suite PASS; login/refresh/logout behavior tests PASS.
UNVERIFIED: token revocation under concurrent logout (no test); production RS256 key rotation procedure.
OPEN FINDINGS / ACCEPTED RISKS: none open.
RESIDUAL RISK: medium, revocation concurrency unproven on a security-sensitive path.
VERDICT: BLOCKED
REQUIRED NEXT ACTION: add concurrent-logout revocation test (critical path); re-run gate. Merge authority remains separate from this assessment.
Connects securely to your tools. The creator never sees your data.
About this skill
Ship Gate is a final go/no-go for a feature you believe is complete. It runs an eight-step gate sequence across scope, build, behavior, runtime, security and operations, then returns exactly one verdict: PASS, PASS_WITH_WARNINGS or BLOCKED. Anything it cannot prove is marked UNVERIFIED instead of being assumed.
The problem it addresses
Ship Gate puts exhaustive verification at the final boundary, not during every early edit. It turns "is this ready to ship?" into a repeatable question: what exactly is being shipped, what was checked, what was not, and what risk remains. A release that rests on unknown critical-path behavior is not downgraded to a warning: the package treats that as a blocker or an incomplete gate.
How it works
It starts with a baseline: the repository, branch or HEAD and comparison base when Git exists, plus the exact release surface and acceptance criteria. If a critical baseline cannot be established, it is marked unverified rather than invented.
Then it applies the gate sequence in order:
- Scope and diff sanity: accidental files, debug output, secrets, generated drift.
- Build quality: format, lint, type and build checks as applicable.
- Behavior: focused tests first, then a broader suite in proportion to risk.
- Rendered and runtime: the critical user path and responsive states where applicable.
- Security and data: auth, secrets, payments, destructive data, migrations and public boundaries.
- Independent assurance: relevant review findings resolved or accepted by authority.
- Operations: configuration, migrations, observability, rollback and deployment notes.
- Evidence accounting: what is verified, what is unverified, and the residual risk.
Verdict rules
BLOCKED is returned for a reproducible critical failure, a known material security or data-loss defect, an unsafe or unknown destructive migration, missing proof for a critical requested path, an unresolved blocker finding, or a delivered surface that materially differs from approved scope. PASS_WITH_WARNINGS is reserved for warnings that are understood, non-blocking and explicitly surfaced.
What you get back
A fixed seven-field report: BASELINE, VERIFIED, UNVERIFIED, OPEN FINDINGS / ACCEPTED RISKS, RESIDUAL RISK, VERDICT and REQUIRED NEXT ACTION. The package includes a filled-in worked example that ends in a BLOCKED verdict and the next action that follows from it.
Boundaries
Read-only by default. A pass is a readiness assessment, not authorization to merge or deploy, and the skill never commits, pushes, installs dependencies or deploys on its own. Project governance documents, accepted ADRs and approved specs outrank the skill. It stops at its handoff boundary and names the next phase instead of certifying downstream work.
How it differs from nearby work
Ship Gate is not a mid-development review: it is meant for the moment a feature is believed complete, not for iterative feedback on work in progress. It is also not deployment automation: it assesses readiness and leaves merge and deploy authority with you.
Evidence discipline
Facts, inferences and unknowns are kept separate, and a passing narrow check counts as proof only of the surface it exercised. The skill does not claim a tool ran when it did not. If a shell, Git, browser, network or MCP capability is absent, it uses a repository-native fallback or marks the item UNVERIFIED or BLOCKED.
Validation scope
The package is five text files: the process, a README, a worked example and two reference documents. It ships no executable scripts, test suite or benchmark results, and this listing makes no claim about measured outcomes or host-specific testing.
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.
Ship Gate produces a readiness assessment, not approval to merge or deploy; that authority stays with you. It is not designed for mid-development review. Its verdict is only as strong as the evidence supplied or gathered: when a shell, Git, browser or network capability is unavailable, the related gates are marked UNVERIFIED. The package ships no executable validators, benchmarks or host-specific test results, and a passing narrow check proves only the surface it exercised.
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 17 days ago
Permissions
Required: read access to the repository files, diffs, test output and project governance documents under review. Optional: running format, lint, type, build and test commands, or checking rendered pages in a browser, when the host provides a shell or browser; without them the gate marks those items UNVERIFIED instead of claiming they ran. No write access, network access, environment variables or external services are required. The skill is read-only and does not commit, push, install dependencies or deploy.