bestmark1

Phased Engineering Pipeline — Testing skill for Claude Code

Testing community

Claude Code skill for building software with AI agents: seven roles, vertical-slice phases, deterministic checks before LLM review, and QA that verifies every acceptance criterion inside a clean check.

How to install Phased Engineering Pipeline

This entry records only its repository, not the path inside it, so there is no exact command to give. Open bestmark1/Phased-Engineering-Pipeline and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Phased Engineering Pipeline does

Claude Code skill for building software with AI agents: seven roles, vertical-slice phases, deterministic checks before LLM review, and QA that verifies every acceptance criterion inside a clean checkout — so 'it works' means it works outside the agent's session.

Alternatives in Testing

  • Factory — Dark Factory Mode - Spec-in, software-out autonomous pipeline 2.8k ★
  • AI Contribution Skills (Engram) — branch-pr: clean branch and PR workflow 2.5k ★
  • Spec Kitty — Spec-Driven Development for serious software developers 1.6k ★

README

phased-engineering-pipeline

Claude Code skill that orchestrates seven specialized roles — eight when entering an existing codebase — through a full BMAD engineering pipeline: **Product → Consistency → Architect → Tech Lead → Developer** with SOLID and SRE code reviewers, then **QA & release verification in a clean checkout** and a **session-archaeology retro**. A role exists only where it must not be the author of what it judges. Includes auto git commits, feature branch workflow, CI/CD awareness, `SPEC_PLAN/` artifact hub, knowledge base (`docs/`) with a surprises log, AGENTS.md project map, decision logs, and tech debt tracking.


What it does

Instead of asking Claude to "build a system" and hoping for the best, this skill enforces a professional BMAD engineering workflow with explicit approval gates:

[Archaeology] → archaeology-report.md (brownfield mode) → ⛔ READ-ONLY COMPLETE
[Product] → Narrative.md + MRD.md (Full mode) + PRD.md → ⛔ USER APPROVAL
  → [Consistency: product] → ⛔ CONSISTENCY PASS
  → git: create feature/{slug} branch, commit product artifacts
  → [Architect] → ARCHITECTURE.md + CONSTITUTION.md + AGENTS.md + docs/ → ⛔ USER APPROVAL
    → [Tech Lead] → IMPLEMENTATION_PLAN.md (with Decision Log) → ⛔ USER APPROVAL
      → [Consistency: full] → cross-artifact-analysis.md → ⛔ CONSISTENCY PASS
      → loop per vertical slice:
          [Developer] → self-review loop → deterministic gate (build/lint/typecheck/test)
          → subtraction pass (what can this phase drop?)
          → [Reviewer SOLID] ‖ [Reviewer SRE]  (topology by review depth)
          → critic loop on findings → APPROVE → next slice
      → [QA & Release] → clean checkout starts → criteria exercised inside it
          → QA PASS + RELEASE READY → [Retro]
          → ⛔ OWNER PERMISSION → git: push → gh pr create → ⛔ USER REVIEWS DIFF

**Domain Analysis (optional)** — A Domain Analyst researches the problem space, stakeholders, competitors, and risks. Asks 5-8 clari