AUDiaGentic

Multi-agent workflow orchestration system for AI coding agents. Provides the infrastructure to plan, execute, and track software engineering work across coordinated agent sessions.

Quick links

• docs/README.md — documentation map and authority rules
• docs/planning/ — planning system (requests, specs, tasks, plans, work packages)
• docs/releases/ — current release and audit docs
• docs/examples/ — example project scaffold

Source code

Installation

For a released build, install the package from PyPI:

python3 -m venv ~/.venvs/audiagentic
source ~/.venvs/audiagentic/bin/activate
python -m pip install --upgrade audiagentic

The GitHub Release workflow builds and attaches the wheel and source archive to each release and publishes them to PyPI using GitHub Trusted Publishing. See docs/releases/PYPI_PUBLISHING.md.

Docker test path

Use the existing Docker base image as the normal test path. Do not rebuild test images for routine validation unless the image inputs changed.

Normal path:

docker run --rm \
  -v "${PWD}:/app" \
  -w /app \
  -e AUDIAGENTIC_DOCKER_TESTS=1 \
  -e AUDIAGENTIC_REPO_ROOT=/app \
  audia-test-base:latest \
  bash -lc "python3 -m pip install --no-cache-dir --break-system-packages -e . pytest pytest-asyncio mcp==1.27.0 && pytest tests/integration/release tests/e2e/release -q"

Rebuild only when one of these changed:

• tests/docker/Dockerfile.test-base
• tests/docker/Dockerfile.test
• tests/docker/Dockerfile.release-test
• image-level package/tool bootstrap requirements
• the local image is missing or known-bad

Use tests/docker/Dockerfile.release-test only when validating the wheel-installed release path or package-data bundling. It is not the default recheck path.

Testing

• See tests/TESTING.md for current test commands, tiers, markers, and architecture notes.