Agent Loop docs
Installation
based-models-agentloop requires Python 3.12 or newer. Install the distribution named
based-models-agentloop; import it in Python as agentloop.
For the core agent loop and its offline FakeClient, no provider extra is needed:
python -m pip install based-models-agentloop
Choose an extra when you connect to a provider or export telemetry:
| Extra | Adds |
|---|---|
anthropic | Anthropic SDK |
openai-compat | HTTP client for OpenAI, OpenRouter, and Modal/vLLM |
otel | OpenTelemetry SDK and OTLP HTTP exporter |
all | All three extras above |
For example, install one provider extra with either of these commands:
python -m pip install 'based-models-agentloop[anthropic]'
python -m pip install 'based-models-agentloop[openai-compat]'
Add the otel extra if the application exports spans or metrics. Install every integration
with all:
python -m pip install 'based-models-agentloop[otel]'
python -m pip install 'based-models-agentloop[all]'
Check that the package imports in the selected Python environment:
python -c 'import agentloop; print("agentloop ready")'
Output:
agentloop ready
Work from this repository
With Python 3.12 or newer available as python, create a virtual environment and install
the project with its development dependencies:
python -m venv .venv
source .venv/bin/activate
python -m pip install -e '.[all,dev]'
Provider SDKs and the OpenTelemetry SDK are imported only when their integrations are
used. A basic import agentloop works without those optional extras.
Next: Hello World.