Agent channel

Connect Claude to your product

If your product runs Peeve, your users can connect Claude to it. They add your scoped Peeve endpoint as a custom MCP connector in Claude, and Claude receives your product's capabilities as typed tools, with the same confirmation and audit guardrails as the human path, limited to exactly that user's permissions. The agent channel is on every Peeve plan.

How your users connect Claude
  1. 01Turn on the agent channelIn Peeve, the agent channel is available on every plan (Starter gives 5 grants per user, Growth 20, Scale unlimited). Nothing to build: your existing capabilities are what Claude will see.
  2. 02Your user gets their endpointEach user has a personal, scoped endpoint of the form mcp.peeve.ai/<workspace>/u/<contact>, minted per user, revocable by them and visible to you.
  3. 03They add it in ClaudeThe user adds that URL as a custom MCP connector in Claude's connector settings, the same way they would add any custom MCP server.
  4. 04Claude gets your capabilities as toolsClaude discovers your capabilities as typed tools, each stating its effect. Reads run live to answer with current data; writes, like change_plan, are confirmed before they run and written to a signed audit log.
Common questions

How do my users connect Claude to my product?

Your users add their personal Peeve endpoint (mcp.peeve.ai/<workspace>/u/<contact>) as a custom MCP connector in Claude's connector settings. Claude then discovers your product's capabilities as typed tools and can call them on the user's behalf, with writes confirmed first. The agent channel is on every Peeve plan, so no upgrade is needed to enable it.

What can Claude do once it is connected?

Claude gets your capabilities rendered as typed tools, for example change_plan, download_invoice or list_orders, the same registry a human sees as cursor and narration. It can read live data to answer, and take a write action after a confirmation step. It has exactly the connecting user's permissions, never more, and every action is rate-limited and audited.

Is it safe to let users connect Claude?

Yes. Claude acts with exactly the calling user's permissions, each grant is scoped to that user and revocable, and anything that changes state is confirmed and written to a signed audit log. The model requests a capability by name and never holds a credential, so an injected instruction can request a capability but can never compose a raw request or reach a key.

Does connecting Claude cost my users extra?

No. Agent tool calls over MCP are logged but not billed today, and the agent channel is on every Peeve plan. Grants are limited by plan (5 per user on Starter, 20 on Growth, unlimited on Scale). If the billing model ever changes, you get a full billing cycle's notice and never a retroactive charge.

Related: Peeve over MCP, all integrations, the agent channel, developer docs.

Serve your users' agents, not just a chat window. 14 days free. No card. No call.