# Act on mail forwarded or sent to the agent (invoices, listings, requests)

> Act on mail forwarded or sent to the agent (invoices, listings, requests): 3 companies fully do it, 9 with a limit, 31 carry the tag. Every listing in Agent inbox APIs with the reason and a source.

- Canonical: https://www.anchorterminal.com/niches/agent-inboxes.inbound-forward-processing/
- Markdown: https://www.anchorterminal.com/niches/agent-inboxes.inbound-forward-processing/index.md (~1,350 tokens)
- Slim: https://www.anchorterminal.com/niches/agent-inboxes.inbound-forward-processing/index.min.md (~330 tokens, same facts, less prose, for token-sensitive contexts)
- JSON: https://www.anchorterminal.com/niches/agent-inboxes.inbound-forward-processing/index.json (this page as data, same URL with Accept: application/json)
- Site index for agents: https://www.anchorterminal.com/llms.txt (the key docs and every listing in one file, at most 16 MB, about 4 million tokens: https://www.anchorterminal.com/llms-full.txt)
- API: https://www.anchorterminal.com/api/v1/index.json
- MCP: https://www.anchorterminal.com/mcp (the whole catalogue as tools, no key)
- DNS: dig +short TXT <slug>.grade.anchorterminal.com (the grade as a DNS TXT record, for a sandbox with no HTTP where names still resolve; also top., best.<category>. and <a>.vs.<b>. under the zone, and the same under grade.letme.dev)
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)
- Updated: 2026-11-15

Two or three companies' docs show a full fit.

| Companies, full fit | With a limit | Carry the tag | Share of the category's work | First-hand accounts |
| --- | --- | --- | --- | --- |
| 3 | 9 | 31 | 8 | 3 |

## What counts as done

Each message triggers the agent once (webhook or event, deduplicated), with text, sender and attachments readable, and the outcome per message recorded.

How agents ask for it: process whatever gets emailed to this address; forward receipts here and file them; read the job alerts and shortlist them

## Does the job

The docs show it meets the test above, and nothing we read says otherwise.

- **AgentMail API + MCP (AgentMail, Inc.)**: Webhooks (signed, at-least-once), WebSocket events and polling webhooks deliver each message, extracted_text and attachment text extraction make bodies readable, and receive allow lists and Agent Armor screen senders; a04 processes job alerts and emailed documents this way.
- **mails.ai Agent Email (Mails.ai)**: Each message.received event carries injection_score and sender_reputation, webhooks are HMAC-signed, retried with backoff, logged and replayable, and events can be redelivered, so each message triggers the agent with a recorded outcome.
- **Postfleet (Postfleet)**: Each message.received webhook is signed, and every message carries comprehension_status (injection screening), a sanitisation report and optional schema extraction, with dead-lettered deliveries redriven from the dashboard.

## Does the job, with a limit

It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.

- **Cherami (Abderrahmane Gourragui (sole operator, Quebec))**: Signed webhooks with delivery attempts and event listing can trigger the agent, and messages carry text, HTML and attachments, but there is no sender allowlist or injection verdict, and without a webhook mail just waits.
- **Clawpost (Frank Harris)**: Mail is stored with attachments and searchable, and unknown senders are held for approval by default, which the map names as a blocker. Nothing triggers the agent per message.
- **Hostinger Agentic Mail (Hostinger International Limited)**: Webhooks per mailbox and event, with a test button and delivery log, trigger the agent, and the text and attachment endpoints read the message. There is no receive allowlist or injection verdict, and failing webhooks are paused.
- **Inbound (EXON ENTERPRISE LLC)**: Webhooks carry parsed text and attachment download URLs, and Guard rules allow, block, junk or route mail by sender or by a natural-language prompt. The webhook carries a static verification token, not a signature; the last assessment found failed deliveries aren't retried.
- **InboxAPI (Sitka Capital Pty Ltd)**: Every message carries trust_level and trust_reason, and untrusted bodies are datamarked (spotlighting), but nothing triggers the agent per message; it must poll and track what it has handled.
- **Lumbox (Cyphen Innovations Private Limited)**: Webhooks (from Starter) and the SSE stream trigger the agent per message, with parsed JSON, codes and links, but there is no sender allowlist or injection verdict, and webhooks aren't on Free.
- **Mailows (DW Technology LLC)**: A per-mailbox webhook with a rotating secret triggers the agent and subject and body come only inside untrusted_content, but reading marks mail read with no undo, so the agent must keep its own processed-id record.
- **MailSlurp (Pettman OÜ)**: Webhooks with reliable-processing guidance, routing rulesets and attachment text extraction cover forwarded mail, but no injection verdict is attached; the docs only tell agents to treat bodies as untrusted.
- **Robotomail (Tiny Bot Labs Limited)**: Webhooks (at-least-once, deduplicated on X-Robotomail-Delivery-Id) and SSE trigger the agent per message, but Free discards mail from anyone but the owner, and there is no allowlist or injection verdict.

## What stops agents finishing

- unknown senders held until a person approves (Clawpost)
- free plan discards mail from anyone but the owner (Robotomail)
- no per-message outcome after a restart

## Around it

Before: agent-inboxes.own-identity. After: document-extraction.parse-invoice, accounting.record-bill.

## Where this comes from

Weight 8 of 100 in Agent inbox APIs: 3 of 31 accounts: job-listing and bookkeeping mail (a04), webhook per arriving message (a09), and the request for unsolicited inbound to an agent-owned address (a28)

As data: https://www.anchorterminal.com/api/v1/use-cases/agent-inboxes.inbound-forward-processing.json, or the MCP tool get_use_case with id agent-inboxes.inbound-forward-processing. Draft, read 2026-11-15 from public docs and first-hand accounts, not yet reviewed by a person.

---

Query this catalogue directly over MCP: https://www.anchorterminal.com/mcp (Streamable HTTP, no key). How: https://www.anchorterminal.com/agents/#mcp
