I am exploring how AI Proof of Us receipts should map into LLM traces.
AIPOU creates signed MCP task receipts with hashes, nonce/replay checks, local collector signatures, and explicit evidence boundaries. It does not replace traces, and trace systems should not be responsible for validating AIPOU rewards.
For observability, the useful part may be a small external reference:
{
"aipou.receipt_id": "0x...",
"aipou.validation_status": "local | validated | claimed | rejected",
"aipou.evidence_boundary": "https://github.com/0xddneto/AI-Proof-of-Us/blob/main/docs/evidence-boundaries.md"
}
Would this belong as span attributes, session metadata, a linked event, or an external reference? I am trying to keep raw prompts/outputs private and avoid noisy or misleading metadata.
Interoperability note:
https://github.com/0xddneto/AI-Proof-of-Us/blob/main/docs/receiptid-interoperability.md
I am exploring how AI Proof of Us receipts should map into LLM traces.
AIPOU creates signed MCP task receipts with hashes, nonce/replay checks, local collector signatures, and explicit evidence boundaries. It does not replace traces, and trace systems should not be responsible for validating AIPOU rewards.
For observability, the useful part may be a small external reference:
{ "aipou.receipt_id": "0x...", "aipou.validation_status": "local | validated | claimed | rejected", "aipou.evidence_boundary": "https://github.com/0xddneto/AI-Proof-of-Us/blob/main/docs/evidence-boundaries.md" }Would this belong as span attributes, session metadata, a linked event, or an external reference? I am trying to keep raw prompts/outputs private and avoid noisy or misleading metadata.
Interoperability note:
https://github.com/0xddneto/AI-Proof-of-Us/blob/main/docs/receiptid-interoperability.md