The problem
Receipts arrive through email, chat, and manual uploads, and are easy to lose, duplicate, or acknowledge before the underlying record is durable.
An agent can coordinate bounded provider transport, duplicate checks, and host-file verification while leaving custody and accounting judgment with the selected backend and owner.
Safe stopping and recovery
Useful even when the whole path cannot run.
- Stop after a staged candidate when host verification cannot proceed.
- Stop after the external record is verified but before acknowledgement when the mailbox epoch or provider result is uncertain.
- If the workspace, hashes, paths, or finance.json re-read fail, stop before acknowledgement and retain only a safe exception reference.
- If acknowledgement is unknown or fails, preserve the verified external record and transfer id, then retry the same epoch-qualified acknowledgement without re-ingesting it.
Documented follow-on work
The next workflow is conditional, not hidden.
Specialists inside this workflow
Clear roles for each part of the job.
Connected work
Built to use the tools you already have.
Imap