# Agent Delegation Receipts

[3 · Permit](/category/permit/)
Candidate · auto-drafted, unreviewed · confidence 60%

| Agent Delegation Receipts |
| --- |
| Category | [Permit](/category/permit/) |
| Governed by | Individual Internet-Draft submitted to the IETF (Network Working Group); not an adopted working group document. |
| Status | Internet-Draft, Intended Status: Informational; expires 12 April 2027 |
| Phase | draft |
| Implementation | unknown |
| Adoption | unmeasured |
| Version | draft-nelson-agent-delegation-receipts-11 (2026-10-09) |
| Specification | [https://datatracker.ietf.org/doc/html/draft-nelson-agent-delegation-receipts-11](https://datatracker.ietf.org/doc/html/draft-nelson-agent-delegation-receipts-11) |
| Found at | No discovery mechanism described in the available page text; receipts are presented by the agent to a verifier before executing an action. |
| Last checked | 2026-10-18 |
| Freshness | current · checked 2026-10-18 |
| Solves | Delegate a user's authority to an agent, Audit what an agent did, Prove an agent's identity |

Agent Delegation Receipts is an agentic-web standard (document format) at step 3, Permit, governed by Individual Internet-Draft submitted to the IETF (Network Working Group); its own status is "Internet-Draft, Intended Status: Informational; expires 12 April 2027" (draft). It answers: How can a human principal's authorization of an AI agent (scope, boundaries, validity window, and operator instruction hash) be recorded as a signed, verifiable receipt that can be checked before the agent acts? Adoption is unmeasured, last verified 2026-10-18.

[Permit](/category/permit/) Defines the Agent Delegation Receipt, a signed record of the scope, hard boundaries, validity window, and operator-instruction hash that a human principal authorized for an AI agent. Version -11 uses a JWS profile (RFC 7515) over JCS-canonicalized JSON (RFC 8785) and a COSE_Sign1 profile (RFC 9052). It also specifies scope attenuation across multi-agent delegation chains and a revocation registry with cascade semantics. The draft states it is an audit and attribution layer for honest deployments, not a control against a malicious operator.

## Discovery and probe

No discovery mechanism described in the available page text; receipts are presented by the agent to a verifier before executing an action.

```
NONE
parse:
on absent: No site-level support signal is defined, so absence cannot be observed.
```

Not probed live: no-fixed-path. candidate record; probe rule pending

## History
- 2026-10-16 — auto-drafted candidate from exa (https://datatracker.ietf.org/doc/html/draft-nelson-agent-delegation-receipts-11)

## Instances

No instances recorded yet.

## See also

## References

- The Internet-Draft defines the Agent Delegation Receipt as a signed record of what a human principal authorized an AI agent to do, using a JWS profile over JCS-canonicalized JSON and a COSE_Sign1 profile, plus scope attenuation for delegation chains and a revocation registry with cascade semantics. — [https://datatracker.ietf.org/doc/html/draft-nelson-agent-delegation-receipts-11](https://datatracker.ietf.org/doc/html/draft-nelson-agent-delegation-receipts-11) (2026-10-09) REPORTED

[JSON](/data/protocols/agent-delegation-receipts.json) · [Markdown](/protocols/agent-delegation-receipts/index.md)