RDLLM Quickstart

RDLLM is an open-source attribution and royalty layer for AI answers. It shows which sources support an answer, how much visible source usage contributed to the output, and whether the answer can be displayed as grounded.

Run A Local Demo

python -m pip install .
rdllm-operator-doctor
PYTHONPATH=src python3 -m rdllm.cli answer "How should AI prove attribution?"

Look for:

• Sources
• Claim Evidence
• support
• text_match
• payout
• disagreement=passed

Run The Service

export RDLLM_SERVICE_TOKEN="${RDLLM_SERVICE_TOKEN:-rdllm-local-dev-token}"
export RDLLM_SERVICE_TOKEN_SHA256="$(python - <<'PY'
import hashlib
import os
print(hashlib.sha256(os.environ["RDLLM_SERVICE_TOKEN"].encode()).hexdigest())
PY
)"
PYTHONPATH=src python3 -m rdllm.service --config examples/service_config.json

Then call /v1/attribute from curl or from the API client examples.

Verify Before Display

Save the response JSON and copied answer text, then run:

PYTHONPATH=src python3 tools/service_response_verify.py --response /tmp/rdllm-response.json --display-text /tmp/rdllm-display.txt
PYTHONPATH=src python3 tools/source_footer_verify.py --footer /tmp/rdllm-footer.json --display-text /tmp/rdllm-display.txt

Only display the answer as grounded when production_display_ready is true and source_grounding_acceptance is passed.

More: explainer, GitHub start guide, and README.