One decorator. That's the whole integration.

$ pip install foxy-audit

from foxy_audit import audit

@audit(policy="soc2")
def ask_model(prompt):
    return llm.generate(prompt)

The decorator wraps your function transparently. Your code runs exactly as before — same return value, same latency — while Foxy records what happened in the background.

What actually leaves your machine

• A SHA-256 digest of the prompt and the response — never the raw text.
• The token count, model name, timestamp, and the policy tag you set.
• The policy verdict from the judge — clean or flagged, and why.

Built to stay out of your way

• Never blocks: hashing and sending happen off the request path.
• Never throws its own errors: if Foxy has a problem, your function still returns normally.
• Survives outages: records queue locally and sync when the backend is reachable again.