- receipt
- loading…
- tool
- loading…
- outcome
- loading…
- charged
- loading…
- permit
- loading…
- idempotency
- loading…
- issued
- loading…
- key
- loading…
Portable receipt / Public evidence
Verify this receipt yourself.
This receipt is self-issued from a non-sensitive
partner.echo call and was issued on
.
It is a published historical sample: verifying it proves the gateway
path signed this receipt with the published key. It is not customer
evidence, and it is not a test of the deployment running today.
01 / Published artifact
Signed claims from the portable bundle.
Loading the published receipt and matching public-key snapshot…
Keep both files together. The receipt contains the signed bytes; the snapshot contains the matching Ed25519 public key.
Download receipt.json Download trust-keys.json02 / Offline verification
Check the signed bytes fully offline.
Install the repository SDK, then verify the two downloaded files
with networking disabled. Exit 0 means verified, 1 means the
signature is invalid, and 2 means the verifier cannot determine.
Exit 0 proves that the receipt matches the supplied key snapshot;
it does not independently authenticate who supplied that snapshot.
Pin the key through a trusted channel before relying on issuer
identity. --expect-issuer checks the claimed origin as
an additional consistency guard.
b2a in the package and command names is short for
business-to-agent — it is the SDK's name, not a separate product.
The verifier ships inside the public source repository. The commands below run as written; the receipt and key snapshot above stay downloadable by anyone.
git clone --depth=1 https://github.com/PetrefiedThunder/agent-middleware-api.git
python -m pip install "./agent-middleware-api/b2a_sdk[verify]"
b2a-verify-receipt \
--bundle receipt.json \
--keys trust-keys.json \
--expect-issuer https://api.thisisatest.tech
03 / Bring your own action
Test the same loop on one MCP tool.
If this evidence shape maps to a real authorization, retry, or audit problem in your stack, email the tool scenario: the tool or action, what goes wrong when it is retried, and how you currently check whether it ran. Include the cost of one duplicate or unproven call, the budget owner, and a decision date. Synthetic or redacted examples only. The pilot is paid, bounded to one tool, and priced after fit is established, with setup and recurring service quoted separately in writing. A call happens only when a scenario needs one. Read the pilot scope and estimate economic fit.