robertbarclayy/conducty/conducty-ship
Claude Code and agent skill by robertbarclayy/conducty
- community
- ext:[{"type":"hook","description":"Pre/post lifecycle hooks"}]
Checking that the SKILL.md file is still in the repository.
Pre-merge / pre-deploy gate. Runs the full ship-readiness battery — lint, typecheck, test suite, secrets scan, dependency-vulnerability check, [[conducty-code-review]] verdict — and writes a `Ship Reports/Ship Report YYYY-MM-DD HHmm.md` note with a single-word verdict. Use when the user says "ship it", "is this ready", "ship gate", "pre-merge check", "ready to merge", or after [[conducty-code-review]] passes.
Install
npx vskill@latest install robertbarclayy/conducty/conducty-shipSemi-extensibleMentions customization, but needs skill-specific knowledge. Read the extensible skills standard
SKILL.md
Conducty Ship — The Pre-Merge Gate
[[conducty-checkpoint]] gates a parallelization group inside a plan. [[conducty-code-review]] gives a holistic human-style read of the branch. conducty-ship is the last gate before merge: it runs the actual mechanical checks every project must pass and produces a single ship-or-don't verdict.
[!important] Read [[conducty-obsidian]] firstOutput goes to the vault asShip Reports/Ship Report YYYY-MM-DD HHmm.md, linked from the plan note and the code-review note.
When to Use
- After [[conducty-code-review]] verdict is
passorneeds-fix(with fixes applied) - Before merging to main, opening a release PR, or deploying
- The user says "ship it", "ready to merge", "ship gate", "pre-merge check"
Skip when:
- The plan hasn't been reviewed yet — run [[conducty-code-review]] first
- The branch is not yet a candidate for merge (still mid-plan)
The Battery
Six gates. Every gate must pass for ship verdict to be `green`. A single failure in any gate produces verdict red. Yellow exists only for Code review: needs-fix with all fixes applied but not re-reviewed — flag it but treat operationally as red.
| # | Gate | What runs | Pass criterion | |
|---|---|---|---|---|
| 1 | Code review | Read Code Reviews/Code Review YYYY-MM-DD HHmm.md from the vault | Verdict pass, or needs-fix with all Important/Critical findings checked off | |
| 2 | Lint / format | Project's lint command (npm run lint, cargo clippy, ruff, etc.) | Exit 0, no warnings the project treats as errors | |
| 3 | Type check | Project's typechecker (tsc --noEmit, mypy, cargo check) | Exit 0, no errors | |
| 4 | Test suite | Full test suite (npm test, cargo test, pytest, etc.) | Exit 0, no failures, no skipped tests added in this branch | |
| 5 | Secrets scan | git diff $(git merge-base HEAD main)..HEAD piped through a secret-pattern grep | No matches against secret patterns (API keys, tokens, credentials) | |
| 6 | Dependency vulnerabilities | npm audit, pip-audit, cargo audit, osv-scanner, etc. — whichever fits the project | No new High/Critical vulns introduced by this branch's dependency changes |
Workflow
Step 1: Read the Plan + Code Review
From the active plan note, find the wikilinked Code Review YYYY-MM-DD HHmm note. Open it. If verdict is block, stop — the branch is not ship-eligible. If verdict is needs-fix, confirm every Critical/Important finding has been addressed (checkbox checked) before proceeding.
Step 2: Resolve Project Commands
Pull the project's commands from [[Context {Project} Tests]] and [[Context {Project} Conventions]] (see [[conducty-context]]). Required entries:
- Lint command
- Typecheck command (or "n/a" with justification — e.g. dynamically typed and the project doesn't run a checker)
- Test command
- Vulnerability check command
If any are missing, ask the user, run them, and ask the user to commit the answer to the project's context note before proceeding (so the next ship doesn't re-ask).
Step 3: Run the Battery
Run gates 2-6 in parallel where possible (Bash with && and &, or sequential with concise output). Capture for each:
- Command run
- Exit code
- Pass/fail
- For failures: the exact failing line(s) — copy verbatim, don't paraphrase
For the secrets scan, use a conservative pattern set:
Tune patterns to the project. False positives are acceptable — a human will check; false negatives are not.
Step 4: Compute Verdict
| Verdict | Condition | |
|---|---|---|
| green | All 6 gates pass. Ship. | |
| red | Any gate failed. Do not ship. | |
| yellow | All mechanical gates pass, but code-review is needs-fix with fixes applied but not re-reviewed. Re-run [[conducty-code-review]] before merging. |
Step 5: Write the Ship Report
Step 6: Wire Wikilinks
- Edit the plan note's
## Review Notessection to add[[Ship Report YYYY-MM-DD HHmm]]. - Edit the code-review note's
## Relatedsection to add the same. - If verdict is red, prepend an entry to
[[Failure Patterns]]describing which gate failed and the leverage point ([[conducty-debug]]). Mechanical-gate failures are usually code-level; secrets-scan / vuln failures are usually plan-level (the prompt should have prevented them).
Step 7: Hand Off
- green: report verdict to user. Do not push, merge, or deploy autonomously — those are user-confirmed actions per Claude Code's blast-radius rules. Suggest the next command.
- yellow / red: present the failures and proposed next plan or fix prompts.
Principles
- Mechanical, not subjective — ship runs commands and reads exit codes. Subjective review is [[conducty-code-review]]'s job.
- All-or-nothing verdict — green/red, not 4-out-of-6. Yellow exists only for the post-review-fix re-run case.
- Failure detail is verbatim — copy the failing command output, never paraphrase. The user (or the next fix prompt) needs to see exactly what broke.
- Ship never auto-merges — verdict is advisory. The user runs
git merge/gh pr merge. - Every red gate is a learning — failures append to
[[Failure Patterns]]so the next plan's prompt-quality gate can prevent the smell that caused them. - Battery is project-specific but consistent — six gates always run; the commands come from the project's context notes.
Security
- Security scanPassed, 95 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/conducty-ship/SKILL.md
Popularity
- Repository stars
- 213
- Repository forks
- 17
- 7-day trend score
- 27
Stars and forks are for the whole GitHub repository, not this skill alone. Refreshed 14 May 2026, 5 months ago.
Extension points
hookPre/post lifecycle hooks
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/robertbarclayy/conducty/conducty-ship)