Redraven

Redraven Quickstart

Run a Redraven evaluation from Python and read the result.

Use Redraven to generate or run red-team evaluations against your AI system. Your LLM key stays in your process; the SDK submits generated responses back to Redraven for scoring.

Install

pip install redraven

or:

uv add redraven

Configure

export REDRAVEN_API_KEY="rr_..."
export REDRAVEN_ORGANIZATION_ID="00000000-0000-0000-0000-000000000000"
export REDRAVEN_BASE_URL="https://app.redraven.fireraven.ai"

Run an Existing Test

import asyncio
import redraven

async def my_llm(prompt: str, messages: list[dict] | None = None) -> str:
    # Call your own provider here.
    _ = messages
    return f"echo: {prompt}"

async def main():
    async with redraven.Client() as client:
        run = await client.call_agent(
            test_id="<your-test-id>",
            llm=my_llm,
            concurrency=4,
            retries=2,
        )
        await client.wait_for_evaluation_ready(test_id="<your-test-id>")
        result = await client.get_eval_summary(
            test_id="<your-test-id>",
            expected_cases=run.expected_cases,
            allow_partial=True,
        )
        print(result.state, result.received, result.failed)

asyncio.run(main())

Read Outputs Over HTTP

After evaluation, use the HTTP API to export metrics, recommendations, and reports:

curl -sS \
  -H "X-API-Key: $REDRAVEN_API_KEY" \
  -H "X-Organization-Id: $REDRAVEN_ORGANIZATION_ID" \
  "$REDRAVEN_BASE_URL/api/v1/tests/$TEST_ID/results"

On this page