actionproof 0.0.3

Tamper-proof audit trail for AI agents — the verifiable layer of agent observability. Signed, offline-verifiable receipts for every agent action. Zero backend.

ActionProof (Python)

Verifiable receipts that prove what your AI agent did. Sign locally, verify anywhere, zero backend. Receipts are cross-compatible with the TypeScript library — a receipt signed in one verifies in the other.

from actionproof import attest, verify, generate_keypair

agent = generate_keypair()

receipt = attest(
    agent,
    type="email.send",
    summary="Sent renewal quote to [e-mail]",
    params={"to": "[e-mail]", "amount": 4200},  # hashed, not stored in clear
    result={"smtp": 250},
    outcome="ok",
)

verify(receipt)  # -> VerifyResult(valid=True, agent="did:key:z6Mk...")

Auto-emit receipts

Decorator — every call to the function emits a signed receipt:

from actionproof import attest_action

@attest_action(agent, type="email.send", on_receipt=store)
def send_email(to, body): ...

LangChain / CrewAI — attest every tool the agent runs, no per-tool code:

from actionproof import ActionProofCallbackHandler

handler = ActionProofCallbackHandler(agent, on_receipt=store)
agent_executor.invoke(input, config={"callbacks": [handler]})

Install & test

pip install -e ".[dev]"
pytest tests/          # 7 tests incl. a TS↔Python cross-language interop check

Requires cryptography. The optional langchain extra makes the callback handler a real BaseCallbackHandler; without it, the handler still works as a plain object.