Fakorede/multi-agent-systems

A monorepo of small, independent agents, each built around a different multi-agent architecture pattern.

★ 0Forks 0PythonGitHub ↗Compare

README

multi-agent-systems

A monorepo of small, independent agents, each built around a different multi-agent architecture pattern. Each agent is a self-contained Python package (its own pyproject.toml, source, README, .env) living in a uv workspace, so they share one resolver and lockfile but stay independently runnable.

Agent Pattern Status
troubleshooting-assistant Skills in progress
--- Sub-agents planned
--- Router planned
--- Hand-off planned

Setup

Install uv, then from the repo root:

uv sync

This resolves and installs dependencies for every workspace member into a single shared .venv at the repo root.

Running an agent

uv run --package <agent-name> <agent-name>/main.py
# e.g.
uv run --package troubleshooting-assistant troubleshooting-assistant/main.py

(--package selects which member's dependencies/environment to use, but script paths are still resolved from the repo root — hence the repeated directory name. cd <agent-name> && uv run main.py avoids that if you find it awkward.)

Each agent also has its own README.md with setup/run details specific to it.

Adding a new agent

  1. Create a new top-level directory for it (its own pyproject.toml, src/, README.md, .env.example).
  2. Add its directory name to members in the root pyproject.toml.
  3. Run uv sync from the repo root.

Contributors

Fakorede

Issues