← ClaudeAtlas

feature-impllisted

Research-first feature implementation workflow. Use when building new features or net-new code that should follow existing codebase patterns. Enforces pattern discovery before coding; dispatches to repo-specific standards skills for convention checks.
jamestexas/agents · ★ 2 · AI & Automation · score 58
Install: claude install-skill jamestexas/agents
# Feature Implementation Workflow Implement new features by discovering and following existing codebase patterns before writing code. This skill prevents the most common implementation defects: duplicated utilities, convention violations, infrastructure resource conflicts, and bolt-on design. **Discipline model:** Phases 0 and 3 are RIGID (produce artifacts, gate the next phase). Phases 1 and 2 are FLEXIBLE (adapt to the problem). ## Arguments $ARGUMENTS Arguments should be a problem statement describing what to build. Examples: - "sync users from identity provider to billing system" - "add a webhook handler for build completion events" - "create a scheduled job that reconciles X with Y" If no arguments provided, ask the user for a problem statement before proceeding. --- ## Phase 0: Research (MANDATORY — BLOCKING) **No code is written until this phase produces a research document.** Launch a separate Explore agent (thoroughness: "very thorough") with the problem statement. The agent must complete all substeps and return a structured summary. ### 0.1 Discover existing patterns Find how similar things are built in this codebase. Use whatever code intelligence is available: - **If mache MCP tools are available:** `mcp__mache__search` with keywords from the problem statement, `mcp__mache__find_callers` on framework entry points, `mcp__mache__find_definition` to locate implementations - **Otherwise:** Grep for domain keywords (e.g., `Reconciler`, `Handler`, `Backfill