deploy-prod skill
Deploys the main branch to Vercel production using a clean git worktree and verifies the deployment with Reelier smoke tests.
- GitHub stars
- 70
- Files
- 1
- Last commit
- Oct 9, 2026
- License
- AGPL-3.0
- Allowed tools
- —
What does the deploy-prod skill do?
deploy-prod deploys the main branch to Vercel production using a clean git worktree and verifies the deployment with Reelier smoke tests.
Summary generated from the skill's SKILL.md.
About the deploy-prod skill
Deploying the main branch to Vercel production using a clean git worktree and verifying the results with Reelier smoke tests is what this skill enables. Developers should load this tool after merging code to main to bypass the unreliable Vercel git integration. It automates a sequence that fetches the latest commit, creates a temporary worktree in the system temp directory, copies the .vercel configuration, and executes the production deployment command via npx vercel deploy.
Once the deployment is live, the skill verifies the status using npx vercel inspect before running a deterministic smoke test through the Reelier CLI. This approach uses Level 0 replay which consumes zero LLM tokens and provides per-step receipts in the .reelier/runs/ directory. If the deployment involves database changes, the process includes a manual verification step on Neon to confirm migrations via the information_schema. The skill concludes by cleaning up the temporary worktree and pruning the git state.
What an agent can do with it
- 1Automate production deploymentsTrigger a safe deployment of the main branch to Vercel using a clean worktree to ensure environment variables are correctly baked in.
- 2Perform cost-effective smoke testingRun deterministic smoke tests using Reelier to verify the deployment without spending LLM tokens.
- 3Verify database migrationsConfirm that new tables or columns exist in Neon after a deployment has run its migration scripts.
- 4Handle deployment failuresDetermine if a smoke test failure requires a human-like judgment or if the assertion simply needs updating due to copy changes.
Written from the skill's SKILL.md. Every use case is checked against a quote from it.
How to install the deploy-prod skill
Each command copies the skill folder into the client's skills directory. Restart the client or start a new session to load it.Install the deploy-prod skill in Claude Code
~/.claude/skills/deploy-prod/SKILL.md
$ npx degit seldonframe/seldonframe/.claude/skills/deploy-prod ~/.claude/skills/deploy-prodInstall the deploy-prod skill in Claude Code (project)
.claude/skills/deploy-prod/SKILL.md
$ npx degit seldonframe/seldonframe/.claude/skills/deploy-prod .claude/skills/deploy-prodInstall the deploy-prod skill in Codex
~/.agents/skills/deploy-prod/SKILL.md
$ npx degit seldonframe/seldonframe/.claude/skills/deploy-prod ~/.agents/skills/deploy-prodInstall the deploy-prod skill in Gemini CLI
~/.gemini/skills/deploy-prod/SKILL.md
$ npx degit seldonframe/seldonframe/.claude/skills/deploy-prod ~/.gemini/skills/deploy-prodInstall the deploy-prod skill in Claude.ai
npx degit seldonframe/seldonframe/.claude/skills/deploy-prod deploy-prod && zip -r deploy-prod.zip deploy-prod # claude.ai → Customize → Skills → + → Create skill → Upload a skill → deploy-prod.zip
SKILL.md
View on GitHubdeploy-prod — clean-worktree deploy + Reelier smoke
Vercel's git integration does not reliably auto-deploy main (verified 2026-07-18), and deploying from the dirty local checkout is forbidden. This is the canonical sequence.
Steps
- Get the merge sha:
git fetch origin mainand note the commit to deploy. - Clean worktree deploy (env vars are baked at build, so this also picks up flag changes):
git worktree add "$env:TEMP\sf-main-deploy" <sha> Copy-Item .vercel "$env:TEMP\sf-main-deploy\.vercel" -Recurse -Force Set-Location "$env:TEMP\sf-main-deploy"; npx vercel deploy --prod - Verify:
npx vercel inspect <deployment-url>→target production+status Ready. - Smoke with Reelier (default — NOT the smoke-runner agent):
Level-0 deterministic replay: ~1-3s, 0 LLM tokens, per-step receipts inSet-Location C:\Users\maxim\CascadeProjects\reelier node dist/cli.js run skills/sf-post-deploy-smoke.skill.md.reelier/runs/. Exit 0 = deploy smoked. This replaces a ~35-60k-token smoke-runner agent dispatch. - Escalation only on divergence: if a step fails, do NOT immediately re-run — a failed
sentinel usually means the page copy changed (update the skill's assert) or the deploy
is genuinely broken (check
npx vercel logs). Dispatch the smoke-runner agent ONLY when the reelier failure needs judgment (e.g. deciding whether copy drift is intentional). - If the deploy touched migrations: verify on Neon that the new table/column exists
(
information_schema) — the build runsdb:migrate, but confirm, don't assume. - Clean up:
git worktree remove "$env:TEMP\sf-main-deploy" --force; git worktree prune.
Why Reelier-first
Deterministic replay with assertions IS the smoke test: same checks every time, receipts
written, zero spend. The agent is the escalation path, not the default — the same
L0-first/escalate-on-divergence ladder the replay product itself uses. When the smoke
skill's sentinels drift because marketing copy changed, updating the assert line in
reelier/skills/sf-post-deploy-smoke.skill.md is the fix (and gets a changelog line).
Files in this skill
1deploy-prod skill questions
How do I install the deploy-prod skill in Claude Code?
Run npx degit seldonframe/seldonframe/.claude/skills/deploy-prod ~/.claude/skills/deploy-prod. Claude Code loads skills from ~/.claude/skills (all projects) or .claude/skills (one project) and uses a skill when its description matches the task.
Which clients can use the deploy-prod skill?
Any client that reads the Agent Skills format: Claude Code, Codex, Gemini CLI and claude.ai (upload as a ZIP). Scripts in the skill run only where the client allows code execution.
Why should I use this skill instead of Vercel's git integration?
Vercel's git integration does not reliably auto-deploy the main branch.
How do I verify if the deployment was successful?
You can verify the deployment by running 'npx vercel inspect <deployment-url>' and checking that the target is production and the status is Ready.
What should I do if the Reelier smoke test fails?
Do not immediately re-run; a failure usually means the page copy changed or the deployment is broken. Check Vercel logs or update the skill's assertion if the change was intentional.