The toolkit targets Python 3.12+ and has a small, well-known dependency set. The core install is fully offline — no API key and no network access are required to analyze input.
- Python 3.12 or newer
pip(oruvfor development)- No external services required for the deterministic engine
pip install devops-ai-toolkitThis installs:
- the
devops-aicommand-line tool (see the CLI guide) - the
devops_ai_toolkitPython package (see the SDK guide) - the packaged YAML knowledge base (see Knowledge base)
Core runtime dependencies: pydantic, pyyaml, typer, rich, jinja2, requests.
The HTTP server (FastAPI + Uvicorn) ships as an optional extra so the core stays lightweight:
pip install 'devops-ai-toolkit[api]'Then start it with either:
devops-ai serve --host 0.0.0.0 --port 8000
# or directly
uvicorn devops_ai_toolkit.api.app:app --host 0.0.0.0 --port 8000See the REST API guide.
devops-ai version
devops-ai list # prints the signatures the toolkit knows aboutimport devops_ai_toolkit
print(devops_ai_toolkit.__version__)The toolkit talks to LLM providers over plain HTTP using requests — there is no required
vendor SDK. You only need to set the relevant environment variables:
| Provider | Required variable |
|---|---|
| Anthropic | ANTHROPIC_API_KEY |
| OpenAI | OPENAI_API_KEY |
| Gemini | GEMINI_API_KEY |
| Ollama | OLLAMA_HOST (local) |
Details in Configuration and AI providers.
For contributing, clone the repo and use uv to create an environment with the dev (and
optionally api) extras:
git clone https://github.com/devopsaitoolkit/devops-ai-toolkit
cd devops-ai-toolkit
uv venv
uv pip install -e '.[all]' # api + dev extrasThe dev extra brings in pytest, pytest-cov, ruff, mypy, pre-commit, and httpx.
See the Developer guide and Testing guide.
pip uninstall devops-ai-toolkitIf a command is missing or the API extra won't import, see the Troubleshooting guide.