Repository navigation
Replies: 1 comment 3 replies
|
Implementation update: I separated the fiscal-document path from screening and published a reviewable draft reference PR: arjun2075/ucp-transaction-requirements#1 The fiscal-only increment uses the existing vendor-extension and full-Order-snapshot model, adds a durable issue/cancel/replace ledger with atomic revisioning, authenticated mutation/retrieval boundaries, tamper checks, concurrent-writer rejection, and exact lost-response retry, and passes 36 local tests against UCP main at ba9795a. I did not open a Core EP because this audit did not establish a missing Core primitive or independent adoption need. It does not claim live screening, tax-authority/SAT/PAC certification, legal compliance, or production transport validation. The useful review question is whether this belongs as a reference integration/implementation guide and what existing fiscal work it should align with. |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
What is the intended UCP integration pattern when a transaction requires jurisdiction-specific verification or fiscal documents?
Examples include counterparty screening and certified electronic invoices. These have different responsibilities and lifecycles, but both raise questions about how agents discover required workflows, obtain trustworthy results, and determine which operations may proceed.
I investigated whether existing UCP mechanisms could support these workflows before proposing new core fields.
Reference implementation
Repository: https://github.com/arjun2075/ucp-transaction-requirements
Reviewed implementation: c81179f
The local prototype defines two experimental extensions:
These are illustrative extension names and proposed contracts, not official UCP capabilities.
Scenarios exercised
The Business requires a verification workflow under its local policy, but the Platform does not support a compatible extension version.
The prototype keeps completion blocked. The extension’s absence from the negotiated capability set does not waive the requirement.
This is implementation policy, not a proposed core required_extensions field.
A signed verifier result covers the buyer, seller, payer, destination, goods, currency and total.
Changing the destination or payer invalidates the accepted result under the experimental extension contract. The Business surfaces fresh verification work and reevaluates completion only after accepting a valid result.
The prototype also rejects tampered signatures, untrusted providers, expired results and results for a different task.
A hypothetical workflow requires certification before Checkout completion. Its signed result identifies a document and content hash; initial document delivery must match that evidence.
After completion, the fiscal-document extension carries document delivery, cancellation and replacement records in full Order snapshots. Document cancellation is distinct from canceling the purchase.
Mexico’s CFDI system motivated this example, but the XML fixtures are explicitly not CFDIs. The prototype does not assume that every jurisdiction or transaction requires certification before payment.
Findings and limits
The example composes with the existing draft Checkout and Order schemas without modifying upstream core schemas.
All 23 local integration tests pass. Capability declarations and response payloads are validated using JSON Schema Draft 2020-12 against UCP revision ad9950b.
Important boundaries:
An extension defines the integration contract; participating services perform the work and enforce the effects they control. Negotiating an extension is not itself evidence that a transaction satisfies its requirements.
Guidance requested
Related discussion
Discussion #926 focuses on counterparty restrictions and screening evidence. This discussion asks about the broader integration pattern while keeping screening and fiscal-document responsibilities distinct.
The prototype demonstrates local feasibility. It does not establish a protocol bypass or a need for a new core compliance layer. I would be happy to narrow it into a documentation contribution based on maintainer guidance.
All reactions