CtrlK
BlogDocsLog inGet started
Tessl Logo

devops-deploy

CI/CD and release engineering router: pipelines, environments, deployment gates, rollbacks, git hooks, watchers, and repo hygiene. Routes one topic per call. Topics: ci-pipeline-design, dependency-management, deployment-checklist, environment-management, file-watcher, git-hook-integration, github-actions-ci, lighthouse-ci, linting-formatting, rollback-strategy.

SKILL.md
Quality
Evals
Security

devops-deploy

Router skill for CI/CD and release engineering. Resolve topic, then Read EXACTLY ONE playbook from routes:. [DOC]

When to use

A delivery / release / repo-hygiene task that maps to one routes: topic. This router owns only these ten topics; hand non-deploy parts to sibling skills. [INFERENCE]

Inputs → Outputs

  • In: topic (required; infer), depth (quick|deep). [CONFIG]
  • Out: the artifacts the playbook defines (workflow YAML, hook script, checklist, env matrix, rollback runbook), each claim tagged. [DOC]

Routing procedure

  1. Map the request to one topic. Synonyms: "release/ship" → deployment-checklist; "GHA/workflow" → github-actions-ci; "revert/undo deploy" → rollback-strategy; "pre-commit/husky" → git-hook-integration; "perf budget/web vitals in CI" → lighthouse-ci; "lockfile/upgrade/audit" → dependency-management. [INFERENCE]
  2. Ambiguous or genuinely multi-topic → ask which one; never guess across. [DOC]
  3. Read that one playbook. deep → apply exhaustively, verify each step; quick → essentials only. [CONFIG]

Spine: Discover → Analyze → Execute → Validate. [DOC]

Gate tooling: assets/ holds the quality rubric and reviewer checklist that back the validation gate below. [DOC]

Validation gate (done = all true)

  • Exactly one topic resolved and one playbook Read — not the cluster. [DOC]
  • Topic ∈ the routes: enum (no invented topics; enum is source of truth). [CONFIG]
  • Script-first honored: prefer scripted/idempotent steps over manual ones. [DOC]
  • Evidence tags present (Alfa core, one spelling); constitution v6.0.0 holds. [DOC]

Anti-patterns

  • Loading multiple playbooks "to be safe" — defeats the router. [INFERENCE]
  • Answering a deploy question inline without Reading the playbook. [INFERENCE]
  • Renaming/extending topics, or editing routes: to fit the request. [CONFIG]
  • Marking done on a green pipeline alone — green ≠ verified behavior. [DOC]

Self-correction triggers

A step lacks its verification → re-read the playbook. Mid-task the real need is a different topic → re-route. [INFERENCE]

Repository
JaviMontano/jm-adk-beta
Last updated
First committed

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.