> ## Documentation Index
> Fetch the complete documentation index at: https://docs.proofable.me/llms.txt
> Use this file to discover all available pages before exploring further.

# Proofable for MCP

> Give AI agents identity, scoped access, trusted context, and verifiable actions through MCP.

Connect one endpoint and every client reads the same profile: the proofs you already hold, the permissions each agent has, and the limits you set. Cursor, Claude, Codex, and VS Code are shortcuts.

Connect once. The same agents, selected context, limits, and current proofs are there in every supported client.

Add `https://mcp.proofable.me/mcp` to the client, then sign in.

Interactive hosts use OAuth. For servers, CI, and headless agents, send a server key as a Bearer token. See [Add Proofable to any MCP client](./setup).

Use a marketplace plugin or registry listing when the host offers one. `npx -y @proofable/sdk setup` writes the same URL from a terminal.

Then ask:

> Show my Proofable profile and current proofs.

Proofable answers **Passed**, **Action needed**, or **Blocked**.

## Next

After you connect, load your profile. Create or import an agent, or publish a listing.

> Show what my agents are allowed to do.

Ask for a dedicated spend account only when the agent should pay from its own account.

Open **Connections** on [proofable.me](https://proofable.me) to link apps. [Connections](../connections)

To sell: set payouts, then publish a listing. Full steps: [Add Proofable to any MCP client](./setup).

Chat on Proofable, or start a [job](/jobs) for background work.

Any tool can connect to this endpoint.

**Private cloud.** Point your server at the same URL. [Run Proofable in a private cloud](../deployment/private-cloud).

Call `proofable_context` once after you connect. Per-tool names, fields, and order: [MCP tools](./tools).

## Privacy

Proofable returns the result and the proof. It does not return the private data used to complete it. Proofs are private by default. The client or runtime must enforce that result.

What supported clients can read when authorized:

* Your profile, agents, and current permissions
* Approved tools, skills, and selected context
* Connected apps and custom MCP servers
* On-demand and scheduled jobs
* Completed checks that are still valid
* Proofs another connected product can confirm
* Vault context when you allow it

To load an agent into a project, see [Connect agent context](../agents/runtime-mount). For a laptop, server, or confidential VM, see [Run Proofable in a private cloud](../deployment/private-cloud).


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.