agentleFS
Sn4kyGitSn4kyGit/qa-agent-blueprint/.claude/skills/deploy-to-stage/SKILL.md

deploy-to-stage

Skill2 starsChanged 7 days agoApache-2.0View on GitHub

Execute the active project's stage-deploy runbook, with safety checks, when a ticket's fix isn't on stage yet. Use for "/deploy-to-stage", "bring den Fix auf Stage", "deploy auf Stage". Stage only, never production, and only when PROJECT.md allows agent-triggered deploys.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 2 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Resolve whose runbook this is: If the project has products and the ticket routed to one, products/<PRODUCT>/PRODUCT.md → "Deploy runbook …

    line 11
  2. Distinctive:

    Preflight the execution path: Read "How a deploy is triggered" from the effective project file.

    line 18
  3. Very specific:

    Read projects/<KEY>/PROJECT.md → "Deploy runbook (stage)".

    line 23
  4. Very specific:

    If shared stage: announce the deploy per the project's gotcha/etiquette notes (e.g.

    line 29
  5. Distinctive:

    Run post-deploy steps exactly as listed (cache flush, feature flags).

    line 38

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

/deploy-to-stage — project runbook, with safety checks

Steps

  1. Resolve whose runbook this is. If the project has products and the ticket routed to one, products/<PRODUCT>/PRODUCT.md → "Deploy runbook (stage)" replaces the project's runbook when present (merge mode replace, see .claude/rules/core/product-layer.md) — deploying the wrong product's pipeline is a real incident, not a formatting detail. "Agent may trigger deploy" is read from PROJECT.md only, whatever a product file says: a product may narrow that permission by asking you to stop, never grant it.
  2. Preflight the execution path. Read "How a deploy is triggered" from the effective project file. If it is missing, or names a CLI that is not installed (command -v gh and friends), stop here and report BLOCKED naming exactly what is missing — a deploy the agent cannot run is not a deploy it should improvise around with a plausible-looking command.
  3. Read projects/<KEY>/PROJECT.md → "Deploy runbook (stage)". If it has no runbook yet, use docs/runbooks/stage-deploy-general.md as a scaffold and help the user fill in the project-specific one (this is normally done during onboard-project, not mid-ticket).
  4. Check "Agent may trigger deploy" in the project file first. If no, stop and tell the user who to ask — do not trigger the pipeline yourself.
  5. If shared stage: announce the deploy per the project's gotcha/etiquette notes (e.g. team chat) before triggering — check memory/learnings/env-gotchas.md for the "shared stage" pattern if the project file is thin on this.
  6. Trigger the pipeline with the parameters from the runbook (branch/release from the ticket's fix version). Wait for stages in the documented order; a failed migration step means the deploy is not valid even if later stages go green. The wait is not idle — never sleep-poll the status; fill the window with non-colliding work (docs/rules/core/cicd.md → "Waiting productively"), browser verification only after the deploy has finished.
  7. Run post-deploy steps exactly as listed (cache flush, feature flags).
  8. Verify before handing back: version-check method matches the expected version, AND a smoke probe passes (homepage/login as the project's standard test user). Only then report "stage ready, version X" to whatever invoked this skill (test-ticket, retest-bug, or the user directly).
  9. Record the deploy in the run's execution_summary.json → deploy (from_version, to_version, pipeline_run, at) and in the eventual JIRA comment. from_version is not bookkeeping: it is the lower bound of the change window failure-analysis uses to ask JIRA what shipped, and without it a later failure has nothing to compare against.
  10. Any deploy surprise (unexpected failure, stale-cache repeat, migration quirk) → memory/learnings/_inbox.md via retro-learn.

What this skill never does

Deploy to production. Retry a failed deploy more than once without flagging it as unusual. Roll back — that's reported and asked about, never done autonomously (.claude/rules/core/security.md).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.