summitsdk 0.1.0

Instrumentation layer for multi-agent AI pipelines

MARL

Simulation and evaluation environment for Summit by Cronys.

Structure

sim/ Planner, Worker, Judge agents eval/ Reward model evaluator improve.py Single-pass improvement loop cycle.py Automated multi-cycle runner marlsdk/ summitsdk package source data/ Traces and evaluation results

Usage

from anthropic import Anthropic
from marlsdk.tracer import Tracer
from marlsdk.exporters.local import LocalExporter

exporter = LocalExporter(output_dir="traces/")
tracer = Tracer(exporter=exporter)

client = Anthropic()
wrapped = tracer.wrap_anthropic(
    client,
    from_agent="planner",
    to_agent="worker",
    task_id="task-001",
    round_trip=1
)

response = wrapped.messages.create(
    model="claude-sonnet-4-20250514",
    max_tokens=1024,
    messages=[{"role": "user", "content": "your prompt"}]
)
# trace written automatically to traces/

OpenAI: use tracer.wrap_openai(client, ...) with the same arguments.