# CommSync

> AI-native unified inbox for SMS + email. CommSync merges every text
> message and email into one threaded conversation per contact, layers an
> AI harness on top (inbox triage, a twice-daily brief, agentic search over
> years of history, autonomous reply agents), and exposes the whole thing
> to external tools over MCP and webhooks.

CommSync unifies SMS (Skyetel, JustCall, Twilio) and email (IMAP/SMTP)
into one inbox per organization. Contacts ("People") own multiple channel
"Identities" (phone numbers, email addresses), so a conversation that
starts as a text and continues by email stays a single thread. On top of
that, an AI harness classifies inbound mail, writes a priority-ranked
daily brief, answers natural-language questions over a team's full
message history ("Ask"), and can run autonomous reply agents on
whitelisted channels under human-approved autonomy modes (draft,
supervised, or fully automatic). Everything a human can do in the app —
read threads, send messages, manage contacts, register webhooks — is
also reachable by an external agent through the MCP server below.

## Product

- [Home](https://commsync.ai/): overview, product tour, positioning.
- [Pricing](https://commsync.ai/pricing): plans, seats, usage limits.
- [Security](https://commsync.ai/security): data handling and encryption posture.
- [Contact](https://commsync.ai/contact): reach the CommSync team.

## Docs (knowledge base)

- [Docs home](https://commsync.ai/docs): full knowledge base index and sidebar.
- [Quickstart](https://commsync.ai/docs/quickstart): connect a channel and send the first message.
- [Concepts](https://commsync.ai/docs/concepts): People, Identities, Threads, Labels — the core data model.
- [Threading](https://commsync.ai/docs/threading): how SMS and email fold into one conversation.
- [AI features](https://commsync.ai/docs/ai): inbox triage, classification, the daily brief.
- [Ask](https://commsync.ai/docs/ask): agentic chat over a team's inbox history.
- [Agents](https://commsync.ai/docs/agents): autonomous AI responders for SMS and email.
- [Contacts](https://commsync.ai/docs/contacts), [Merging](https://commsync.ai/docs/merging), [Labels](https://commsync.ai/docs/labels): the contact graph.
- [Roles and permissions](https://commsync.ai/docs/roles-and-permissions): org roles and channel access.
- [API keys](https://commsync.ai/docs/api-keys): mint credentials for MCP or REST access.

## For AI agents and integrators

- [MCP agent reference](https://commsync.ai/docs/mcp.txt): plain-text tool catalog and auth flow for the CommSync MCP server at `https://server.commsync.ai/api/mcp` — fetch this first.
- [MCP human docs](https://commsync.ai/docs/mcp): the same tool catalog, browsable.
- [Webhooks agent reference](https://commsync.ai/docs/webhooks.txt): event catalog, signing, retries — the push-based alternative to polling MCP.
- [Webhooks machine spec](https://commsync.ai/docs/webhooks.json): JSON-Schema-ish spec of the webhook envelope and every event.
- [Webhooks human docs](https://commsync.ai/docs/webhooks): the same reference, browsable.
- [API catalog](https://commsync.ai/.well-known/api-catalog): RFC 9727 linkset indexing the REST API and the MCP endpoint.
- [MCP server card](https://commsync.ai/.well-known/mcp/server-card.json): SEP-1649 server card — transport, auth, and tool catalog as JSON.
- [auth.md](https://commsync.ai/auth.md): how to get an API key, at the well-known auth.md location.
- [Agent discovery](https://commsync.ai/docs/agent-discovery): how all of the above fit together.
- [Sitemap](https://commsync.ai/sitemap.xml): every indexable marketing and docs URL.

## Crawling policy

CommSync explicitly welcomes AI training, AI search, and agent
crawlers across its public marketing site and /docs knowledge base —
see https://commsync.ai/robots.txt for the full allow list (GPTBot,
ClaudeBot, PerplexityBot, Google-Extended, and more, all invited by
name) and its Content-Signal directive (search=yes, ai-train=yes,
ai-input=yes). The authenticated application (inbox, settings, admin)
and the REST/MCP API require a signed-in user or an API key and are
intentionally excluded from crawling: there's nothing to index there,
and some of it is private customer data.

## Markdown on request

Any page under /docs, and this homepage, also answer `Accept:
text/markdown` with the raw markdown instead of rendered HTML —
same URL, `Content-Type: text/markdown`, `Vary: Accept`.
```
curl -H "Accept: text/markdown" https://commsync.ai/docs/mcp
```