agentic-titan

Getting Started

This guide is the short path from a fresh checkout to a useful local development loop for Agentic Titan.

1. Prepare Python

Use Python 3.11 or newer. The CI matrix covers Python 3.11 and 3.12.

python3.11 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
pip install -e ".[dev,dashboard,auth,ratelimit]"

If you use uv, the equivalent local command is:

uv run --python 3.11 --with '.[dev,dashboard,auth,ratelimit]' pytest

2. Run The Quality Loop

Run these before opening a pull request:

ruff check .
mypy .
pytest

For the same coverage mode used in CI:

pytest --cov=. --cov-report=xml --cov-report=term --cov-report=html

3. Work In The Right Layer

The repo is organized as root-level Python packages:

Path Purpose
titan/ CLI, API, orchestration, safety, learning, and workflow code.
hive/ Collective behavior primitives: memory, topology, stigmergy, and experiments.
agents/ Agent archetypes and reusable agent behaviors.
adapters/ LLM provider adapters and routing.
runtime/ Local and isolated execution backends.
dashboard/ FastAPI dashboard and templates.
.ci/ Local governance and quality scripts used by CI.

4. Choose The Smallest Verification Set First

For a narrow fix, start with the relevant test file:

pytest tests/test_hive/test_perceptual_gating.py
pytest tests/ci/test_ai_quality_gate.py

Then run the broader gates before pushing:

ruff check .
mypy .
pytest --cov=. --cov-report=xml --cov-report=term --cov-report=html

5. Respect The Governance Boundary

This repo participates in ORGAN-IV orchestration. Keep dependencies flowing in the approved direction, do not commit secrets, and do not bypass the branch-protection dependency validation status. If a pull request is blocked on validate-dependencies, run the canonical dependency validator and publish a status only after it passes.