atlasclaw
Core runtime: API layer, agent engine, session/memory, built-in tools, workflow orchestration, and the main docs.
View on GitHubUnderstand the repos, then start the core runtime.
AtlasClaw gives teams a practical way to build one multi-user AI agent and conversational execution layer across existing enterprise systems.
Repository layout
The three public AtlasClaw repositories.
atlasclaw-providers
Provider packages, starter patterns, and reference implementations such as Jira and SmartCMP.
View on GitHubatlasclaw-doc
Versioned installation, administration, user, Core, Provider integration, and reference documentation.
View on GitHubatlasclaw-web
The public website for atlasclaw.ai, built as a static Astro site.
View on GitHub Shortest path
Minimal local startup flow.
- Create a Python virtual environment in the `atlasclaw` repo.
- Install the core dependencies from `requirements.txt`.
- Point `providers_root` at the external providers repo.
- Configure an LLM provider in `atlasclaw.json`.
- Start the FastAPI service and open the web UI.
Commands
Core runtime commands.
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
uvicorn app.atlasclaw.main:app --reload --host 0.0.0.0 --port 8000 atlasclaw.json
The key configuration anchor.
`providers_root` is what connects the core repo to the external provider repo.
{
"providers_root": "../atlasclaw-providers/providers",
"model": {
"primary": "kimi/kimi-k2.5",
"temperature": 0.7,
"providers": {
"kimi": {
"base_url": "${ANTHROPIC_BASE_URL}",
"api_key": "${ANTHROPIC_API_KEY}",
"api_type": "anthropic"
}
}
}
}