← ClaudeAtlas

qualitylisted

Autonomous quality loop with configurable thoroughness (95% or 98%). Runs lint, tests, build, security scans, and specialized quality agents. Auto-fixes issues and creates PRs.
buildproven/claude-kit · ★ 0 · AI & Automation · score 75
Install: claude install-skill buildproven/claude-kit
> **FORKED SKILL INVOCATION — EXECUTE STEP -1 IMMEDIATELY.** > > This skill was already invoked by `/bs:quality`. In this forked context, the > active task is the quality workflow defined below; task arguments arrive via > the `--args-file` channel described in Step -1, not necessarily as a fresh > user message. > > If the visible context contains only system reminders or appears to have no > new user prompt, do not wait for another request and do not report "no user > task." Treat that state as expected fork startup context and immediately run > the Step -1 args-file extraction block. # Quality Skill — Autonomous Quality Loop Makes your project ship-ready in one autonomous command. Replaces manual review cycles with parallel quality agents. **CRITICAL: This is AUTONOMOUS. Do NOT stop and ask the user between loops.** > **Completion condition (for `/goal`-driven runs):** > `/goal all tests pass, lint is clean, and type errors are zero — verified by running the full quality suite` > > When invoked via `claude -p "/goal ..."` or in a loop context, the Haiku evaluator > checks after each turn whether the condition is met. On failure it provides a reason > that guides the next turn automatically. This makes CI-mode runs self-terminating > without manual polling. > **NEVER divert to "investigate uncommitted state" mode.** Once Step -1 resolves > a target (PR, branch, worktree path, or cwd-worktree), execute the quality > pipeline (Step 0 → Step 1 → Step 2 → … → merge). Uncomm