Installation¶
AgentLoop requires Python 3.11 or newer. The package is agentloop-profiler and
the Python import and CLI command are agentloop.
Published package¶
Create an environment for your project and install the package.
Check the installed version and create a local example trace.
quickstart records a synthetic example and writes runs/agentloop_quickstart.json.
Continue with your first useful trace.
Published versions are listed on PyPI and GitHub Releases.
Current source¶
The guided improve commands require 0.2.0 or newer. To try development changes,
install the current source into your environment.
For a development checkout with examples and validation scripts, use Git and uv.
git clone https://github.com/dipeshbabu/agentloop.git
cd agentloop
uv sync --locked --dev
uv run --frozen agentloop improve --help
The guided optimization guide walks through a complete offline example and connecting your own Python agent.
Optional components¶
Install the extras your application uses.
| Extra | Included component |
|---|---|
dashboard |
Streamlit dashboard |
server |
HTTP API server |
postgres |
PostgreSQL storage driver |
instrumentation |
OpenAI SDK instrumentation |
all |
All application extras |
Use the dashboard guide or server deployment guide to configure these components. Upgrade notes cover stored data, client compatibility and optional dependency updates.
TypeScript¶
The separately versioned TypeScript SDK supports Node.js 22 and newer. Follow its source installation instructions. Its initial npm publication is pending.