Back to catalog / Agent workflows

Review an invoice and contract batch under AP policy

Delegate invoices and contracts to specialists under one AP policy and produce per-file reports plus a summary; people make every approval.

Goal

Delegate invoices and contracts to specialists under one AP policy and produce per-file reports plus a summary; people make every approval.

Mount synthetic or approved files and the expense-review-policy skill in the sandbox. [src-openai-review-invoice-contract-batch-under-ap-policy]
Generate a separate review report for each file with exceptions and policy references. [src-openai-review-invoice-contract-batch-under-ap-policy]
Summarize batch exceptions for human review; do not initiate payment. [src-openai-review-invoice-contract-batch-under-ap-policy]

Verify one report per input, matching summary counts and exceptions, and confirm no payment action occurred.

If a report is missing, inspect mount paths and that file’s subagent result; do not treat missing output as compliant.

Python 3.14+, uv, Docker sandbox, Agents API key plus restricted executor key, AP policy, and test files.

Sandbox reads mounted files and writes reports; model/sandbox may be billed but does not pay or approve contracts. Keep keys in environment config, not sandbox files.

The example requires Docker and specific API versions; business approval and payment require separate controls.

These values describe the upstream repository, not popularity or likes for an individual skill or playbook.

No metrics snapshot is available yet.

License: MIT