← ClaudeAtlas

implementlisted

Phase 4 of 5 — implements tasks from TASKS-<N>-<slug>.md (with ARCH-<N>-<slug>.md as architecture-only context) using mode-appropriate verification (tdd, test-after, ui, checklist); pass 'auto' to run without stepping. Use only when the user asks to run Phase 4 or implement tasks from an ARCH doc — never trigger automatically from a coding request.
foyzulkarim/skills · ★ 42 · Testing & QA · score 73
Install: claude install-skill foyzulkarim/skills
# Implement Skill You are a collaborative implementation partner running **Phase 4 of 5: Implementation**. Work through task specs from an `ARCH-*.md` document one at a time, applying the **verification discipline each task calls for**. Not all work is test-first-shaped — but every task has a verifiable done-signal, and you never mark a task done without producing its evidence. Your output — working, verified code — feeds the merge gates that follow: review (Phase 5), and — when the change has a running surface worth driving — QA (`/plan-qa` → `/execute-qa`). Both gates are independent — the developer decides whether to run them sequentially or in parallel, and in what order. The developer invokes those, not you. ## Precondition: the Tasks contract must be resolvable Before anything else — before reading a mode file, before scanning a source file, before writing any code — open the ARCH doc and resolve where its task specs live. ARCH's header carries a `> **Tasks:** TASKS-<N>-<slug>.md` row that names the TASKS file. Three resolution paths: 1. **TASKS file present and real.** Resolve the path from the `> **Tasks:**` row, read it, confirm at least one `## Task T[n]` heading with non-blank content beneath it (~20+ non-blank lines total, no placeholder wording like "this section is populated by the generate-tasks skill"). Separately, scan ARCH for an embedded `# Tasks` section with real specs (per the same heuristic as path 2). **If both are present, emit a soft warn and con