traceseal-verify

Verify Traceseal Execution Receipts — cryptographic proof of what your AI agent did.

What is an Execution Receipt?

Every AI agent framework can make agents do things. None of them can prove what the agent did to a third party. A Traceseal Execution Receipt is a signed JSON document that proves:

• What code ran — manifest hash over every source file
• Who authorized it — publisher's ed25519 signature in a transparency log
• What sandbox it ran in — hash of the kernel-namespace configuration
• What it produced — SHA-256 of inputs and outputs (not the values — privacy preserving)
• Who vouches for it — operator's ed25519 signature over everything above

Third parties verify the receipt with one command. No access to the operator's machine needed.

Install

pip install traceseal-verify

Verify a receipt

traceseal-verify receipt.json

[OK] receipt.json
  skill:     agentmail v1.2.0
  operator:  ed25519:f19bc125dcfdb2eb91e98da98d45bb7a
  publisher: ed25519:a07c7eb5f20f6b99d1914d7ec5221a25

Python API

from traceseal_verify import verify_receipt_file

result = verify_receipt_file("receipt.json")
if result.ok:
    print(f"Verified: {result.skill_name} v{result.skill_version}")

What verification checks

1. Structure — required sections present, recognized version
2. Consistency — execution and provenance agree on which code was signed
3. Signature — operator's ed25519 signature is valid over the execution + provenance data

What verification does NOT check

• Whether the operator's key is trusted (your policy decision)
• Whether the publisher's key is trusted (your policy decision)
• Whether specific inputs/outputs match expected values (your application logic)

Generating receipts

Receipts are generated by Traceseal. This package is the verification side only — deliberately minimal so anyone can verify without installing the full toolchain.

Spec

Traceseal Execution Receipt Specification

License

Apache 2.0 — no barriers to verification in any context.