Connect one Proofable account to any MCP client. Your AI can check who is already verified, act within the limits you set, and take on work you can see and stop.
Start
Add the hosted endpoint to your MCP client and finish sign-in:
Then ask:
Show my Proofable profile and the proofs I can reuse.
Proofable returns Passed, Action needed, or Blocked. It reuses a current proof before starting another verification.
Connect your client
What agents can do
- Check whether you already hold the proof a task requires.
- Show the next verification step only when a current proof is missing.
- Load an agent’s current identity, permissions, skills, and selected context.
- Hand work to an agent as a job. The job uses the browser, search and connected apps its permission allows, on your credits.
- Revoke a proof or an agent permission, everywhere at once.
- Read the same authorized profile and proofs across supported MCP clients.
Call proofable_context once after connecting. Use the MCP tool reference for exact inputs and result shapes.
Build with Proofable
Use the JavaScript SDK, HTTP API, or hosted verification to add Proofable to your product. Servers, CI, and headless agents use access keys against the same hosted MCP endpoint.
The Proofable SDK on GitHub includes the CLI and application examples.
Privacy
A check returns its result and proof, and the private data used to complete it stays out of the response. Proofs are private by default. Your client or runtime enforces the returned result and permissions.
When authorized, agents can read:
- The signed-in profile and current agent permissions
- Approved tools, skills, and selected context
- Current proofs and verification status
- Jobs, their status and any step waiting for your approval
To load one registered agent into a project, see Connect agent context. Last modified on November 17, 2026