Runlet Harness

Application integration helpers for Runlet.

runlet-harness keeps third-party adapters, environment-based configuration, and background delivery concerns outside the small Runlet runtime core.

Install

Base package:

pip install runlet-harness

With Langfuse support:

pip install "runlet-harness[langfuse]"

Langfuse

Cloud configuration:

LANGFUSE_PUBLIC_KEY=pk-lf-...
LANGFUSE_SECRET_KEY=sk-lf-...

Self-hosted configuration:

LANGFUSE_PUBLIC_KEY=pk-lf-...
LANGFUSE_SECRET_KEY=sk-lf-...
LANGFUSE_BASE_URL=https://langfuse.example.com

Usage:

from runlet import CompositeEventSink, InMemoryObserver, Runtime
from runlet_harness.observability import LangfuseEventSink

observer = InMemoryObserver()
langfuse = LangfuseEventSink.from_env()

sinks = [observer]
if langfuse is not None:
    sinks.append(langfuse)

runtime = Runtime(event_sink=CompositeEventSink(sinks))

try:
    result = await runtime.run(agent, "hello")
finally:
    if langfuse is not None:
        await langfuse.shutdown()

By default, the Langfuse sink records metadata, status, timing, tool names, and final outputs. It does not record inputs, reasoning, streaming deltas, tool arguments, tool results, or human-submitted values unless explicitly configured.

Development

Run the test suite:

PYTHONPATH=src python -m unittest discover tests

Run type checking after installing development dependencies:

pyright

Release

Runlet Harness publishes to PyPI from Git tags through GitHub Actions.

Typical release flow:

1. Update the version in pyproject.toml.
2. Merge to main.
3. Create a tag such as v0.1.0a1.
4. Push the tag.

git tag v0.1.0a1
git push origin v0.1.0a1