Skip to content
Postcept
Postcept Receipt

What is a Postcept Receipt?

A Postcept Receipt is signed, tamper-evident proof that a high-risk AI-agent action was checked against the system of record. It records what the agent claimed, which operation was verified, what systems were checked, which postconditions passed or failed, and whether completion was verified.

Definition

Definition

A Postcept Receipt is a signed, tamper-evident proof that a high-risk AI-agent action was checked against the system of record and classified as verified, incomplete, duplicated, mismatched, or policy-failed.

It is the artifact at the end of Proof-of-Completion: every high-risk agent action should leave a receipt.

Receipt fields

Each receipt is a structured, signed object with these fields:

receipt ID
Unique, signed identifier for this evidence object.
operation ID
The agent operation that was verified, stable across retries.
agent ID
Which agent claimed the action.
claimed action
The action the agent reported completing.
contract digest
The versioned definition of done this receipt was judged against.
systems checked
Each system of record that was queried.
postconditions evaluated
Every check run: passed, failed, or skipped, with expected and observed values.
result
verified · incomplete · duplicated · mismatched · policy-failed.
lifecycle
Where the action sits in the provider's lifecycle: finalized, pending, reversed, indeterminate.
safe to claim complete
The one boolean a workflow acts on, inside the signed body.
correlation strength
How the claim was located in the system of record.
observation role
For relay-backed verifications: the relay's own signature over what it observed, verifiable separately from Postcept's.
signature metadata
Algorithm, signing key id, and canonicalization scheme, all inside the signed body.
timestamp
When verification ran.
supersedes
The earlier receipt this one replaces when the truth changed. Old receipts stay verifiable forever.
signature
Ed25519 signature over the receipt contents.
Postcept Receipt
Verified completion
Receipt ID
pcpt_rcpt_01HY9Q...
Operation ID
op_refund_8F31
Agent
SupportAgent-04
Claimed action
refund_customer
Systems checked
Stripe, Zendesk
Timestamp
2026-06-23 09:41:26 UTC
Signaturevalid

Result classifications

Every receipt resolves to exactly one result, so recovery workflows can branch deterministically:

Verified

Completion confirmed against the system of record.

Incomplete

The claimed action did not land.

Duplicated

The same action was attempted more than once.

Mismatched

Amount, customer, or status did not match.

Policy-failed

A required verification or approval was skipped.

Receipts for failure and duplication

Receipts are issued whether the action succeeded or not. A completion gap and a blocked duplicate are both first-class, signed outcomes, the evidence your team needs for audit, recovery, and customer communication.

Postcept Receipt
Completion gap detected
Receipt ID
pcpt_rcpt_01HY9X...
Operation ID
op_refund_8F31
Agent
SupportAgent-04
Claimed action
refund_customer
Systems checked
Stripe, Zendesk
Timestamp
2026-06-23 09:41:26 UTC
Signaturevalid
Postcept Receipt
Duplicate completion blocked
Receipt ID
pcpt_rcpt_01HY9R...
Operation ID
op_refund_8F31
Agent
SupportAgent-04
Claimed action
refund_customer
Systems checked
Stripe
Timestamp
2026-06-23 09:41:31 UTC
Signaturevalid

Why signed and tamper-evident

Because receipts become audit evidence, they are signed over their full contents. The signature makes any later modification detectable, and the signed checks show a reviewer, for each one, the value expected and the value found in the system of record. This is how a Verified Completion Rate is backed by proof rather than estimation.

Put a receipt behind every agent action.

Issue a signed, tamper-evident completion receipt for every high-risk action your agents take.