Add ship-gate: Pre-production audit skill - #527
Conversation
alirezarezvani
left a comment
There was a problem hiding this comment.
Hi @rx4u — thanks for this submission. The content quality here is genuinely strong — 89 well-organized checks with actual grep/regex patterns across 8 categories is substantial work, and the pre-deploy intercept design is clever. This is much more than a stub.
However, a few things need fixing before we can merge.
1. "Production-tested" claim
The source repo was created today (April 13) with 0 stars and 0 forks. The PR body says these are "production-tested skills" — we can't verify that. Please either:
- Remove the claim, or
- Provide evidence (screenshots, deployment logs, links to projects where this was used)
2. Non-existent skill references in SKILL.md
Lines 190-196 reference skills that don't exist in our repo:
- app-planner
- subagent-orchestrator
- backend-patterns
- shadcn-stack / heroui-stack
Please replace these with actual skills from our repo that relate. Suggestions:
## Integration Points
- **karpathy-coder**: run ship-gate after karpathy-check passes — simplicity first, then production readiness
- **adversarial-reviewer**: deep security review for items ship-gate flags as critical
- **security-pen-testing**: penetration testing methodology for SEC-category findings
- **code-reviewer**: general code quality review complements ship-gate's automated checks
- **systematic-debugging**: deep investigation of flagged issues3. Missing frontmatter fields
Add license: MIT to the SKILL.md frontmatter. Current frontmatter:
---
name: ship-gate
description: >
Pre-production audit that scans a codebase...
metadata:
author: Rajaraman Arumugam
version: 1.0.0
---Should be:
---
name: ship-gate
description: >
Pre-production audit that scans a codebase...
license: MIT
metadata:
author: Rajaraman Arumugam
version: 1.0.0
---4. Nice-to-have: a Python scanner script
Your patterns.md contains 687 lines of grep patterns that are begging to be a CLI tool. Adding a scripts/ship_gate_scanner.py that runs the auto-detectable checks and produces a pass/fail report would elevate this from "good documentation" to "production-grade skill" and match the quality bar of our other engineering skills (which all ship Python tooling).
This isn't blocking — we'd accept without it — but it would make the skill significantly more useful and we'd be happy to help you build it if you want.
Summary
| Item | Status |
|---|---|
| Content quality | ✅ Strong — 89 checks, real patterns, good structure |
| Production claim | ❌ Fix — remove or substantiate |
| Skill references | ❌ Fix — replace with skills that exist in our repo |
| Frontmatter | ❌ Fix — add license: MIT |
| Python script | ⚡ Nice-to-have (not blocking) |
Looking forward to the updated PR — this is close to mergeable.
- Remove focused-fix from Integration Points (was a substitution for systematic-debugging which does not exist in this repo) - Add scripts/ship_gate_scanner.py: stdlib-only pre-production audit CLI covering all 8 categories (SEC, DB, CODE, DEP, AI, DEPLOY, FE, OBS) with JSON output, ANSI color, interactive manual checks, and exit codes
|
Quality review — the skill itself looks good (MIT, stdlib-only Python, deterministic, no network calls, valid SKILL.md frontmatter, no red flags). Two small things to address before merge: 1. Path rebase. Current 2. Plugin manifest (optional, recommended). Standalone plugins typically include a Once rebased, ping here and we'll get it merged. Thanks for the contribution — the stdlib-only design and 89-check coverage are exactly what we look for. Generated by Claude Code |
|
Hi @rx4u — the ship-gate skill landed on Generated by Claude Code |
Summary
Adds ship-gate, a pre-production audit skill for the `engineering/` category.
Scans a codebase across 8 categories — security, database, deployment, code quality, AI/LLM, dependencies, frontend, and observability — with 89 automated and manual checks. Intercepts deploy-intent phrases ("push to production", "ship it", "go live") and blocks until critical issues are resolved. Stack-agnostic.
Checklist
Type of Change
Testing
Manually tested against a local Next.js + Supabase project. Ran the skill via Claude Code, confirmed:
This is v1.0.0. Includes `scripts/ship_gate_scanner.py` — a stdlib-only Python CLI that runs all auto-detectable checks, with JSON output mode, ANSI color, interactive manual prompts, and exit codes (0=CLEAR, 1=CRITICAL, 2=HIGH).