Installation

Install agentling from PyPI:

pip install agentling

Or with uv:

uv add agentling

Requirements

• Python 3.11 or newer.
• The only runtime dependencies are openai (the client used by the built-in provider adapter) and pyyaml (for skill frontmatter).

Configure a provider key

The built-in OpenAIModel adapter reads your key from the environment:

export OPENAI_API_KEY="sk-..."

You can point the same adapter at any OpenAI-compatible endpoint (a local server, a gateway, or another vendor's compatible API) by passing base_url. See Using other providers.

Verify the install

Run a quick import check:

python -c "import agentling; print(agentling.__name__)"

If that prints agentling, you are ready to build your first agent.

Try the examples

The repository ships runnable examples. The offline ones need no API key at all:

git clone https://github.com/folathecoder/agentling
cd agentling
uv sync
 
uv run python examples/cli_failure_recovery.py   # recover from tool errors
uv run python examples/cli_memory_chat.py        # persist and continue a session
uv run python examples/cli_advanced_observer.py  # watch every event type

The rest use an OpenAI-compatible model. Set OPENAI_API_KEY, and optionally AGENTLING_EXAMPLE_MODEL:

uv run python examples/cli_math_tutor.py               # smallest useful agent
uv run python examples/cli_repo_assistant.py "..."     # streaming plus safe file tools
uv run python examples/cli_notes_agent.py              # file-backed notes tools

Next steps

• Quickstart: build and run your first agent.
• Tools: how functions become schema-validated tools.