wholeloop-cli 0.2.0

Install WholeLoop agent skills into app repos (Cursor, Claude Code, VS Code)

WholeLoop

WholeLoop is an agentic delivery workflow (ADWF): product specs → user stories (Linear, Jira, or manual) → gated agents in the IDE → implementation → review → PR → handoff.

Portable skills + CLI + docs. No background orchestrator, no n8n — developers run the pipeline in Cursor, Claude Code, or VS Code.

What you get

Area	In this repo
Skills	agents/skills/*/SKILL.md
CLI	wholeloop-cli on PyPI → wholeloop init — install/ · docs/CLI.md
Trackers	docs/TRACKERS.md — Linear, Jira, manual
Spec template	references/SPEC.template.md (product repo)
Conventions	references/PROJECT_CONVENTIONS.template.md
Workflow	docs/WORKFLOW_PRODUCT_LINEAR.md

Principles

1. Spec in product repo — ARTIFACT-WAL; stories in your issue tracker (or manual list).
2. One run per delivery — workspace/runs/<run-key>/context.json (+ plan.md per story).
3. Human gates — After spec-review and planner (and PR when applicable).
4. IDE-native — Skills + MCP or pasted stories; no WholeLoop server to deploy.

Quick start

1. Read docs/WORKFLOW_PRODUCT_LINEAR.md.

2. Install the CLI:

   macOS (Homebrew):

   brew install uv
   uv tool install wholeloop-cli
   

   Any OS (uv already installed):

   uv tool install wholeloop-cli
   

   pipx, pip, Git, upgrades: install/README.md.

3. In your app repo:

   wholeloop init
   wholeloop doctor
   wholeloop conventions bootstrap   # or --conventions-from team file
   

4. Run the project-conventions agent in your IDE → approve .agents/skills/references/project-conventions.md.

5. Product repo: copy references/SPEC.template.md → specs/.

docs/SETUP_NEW_PROJECT.md · docs/IDE_SETUP.md · GUIDELINES.md

Version: skills and docs target v0.2. Check PyPI for the latest CLI; use uv tool install wholeloop-cli --upgrade --force after publishing a new release.

License

MIT — see LICENSE.

Contributing

See CONTRIBUTING.md.