HeadMaster

Open-source ADLC (AI-Driven Delivery Lifecycle) built on Claude Code.
Turns a single conversation into a PRD, a TDD, Jira stories, code, tests, and a reviewed PR.

View on GitHub Quick Start Discussions

What it does

HeadMaster is an orchestration layer that sits alongside your feature repositories. Open Claude Code from the HeadMaster directory, point it at a project, and run slash commands to move features through five stages:

/init-feature  →  /plan  →  /design  →  /breakdown  →  /execute  →  PR

Each stage produces structured, human-reviewable artifacts. Nothing merges automatically. Every gate requires explicit approval.

Pipeline

13 specialized agents — requirements, PRD, design, TDD, developer, code review, QA, release, retrospective, plus a research and codebase analyst
19 skills — init-feature, plan, design, breakdown, execute, implement, security-scan, review-code, qa-integration, review-system, jira-ops, reopen, retrospect, publish-confluence, archive-feature, compress, draw, curate-memory, setup-env
Tier auto-classification — XS / S / M / L workflows skip or require stages based on feature size
Subagent isolation — review/QA/TDD-reviewer agents never see implementation context (enforced by a pre-spawn hook)
Failure ledger — per-story retry log that prevents repeating broken approaches
Self-learning retrospective — auto-applies agent memory and pipeline learnings; routes config proposals to a human queue

Get started

git clone https://github.com/munna-chauhan/HeadMaster.git && cd HeadMaster
cp config.yml.example config.yml
python scripts/setup_projects.py
claude
/init-feature "Add PDF invoice export"

Requirements: Claude Code, Python 3.9+, Node.js 18+.