Skip to content

Security: proofable/mcp

Security

SECURITY.md

Security

Report a vulnerability

Email dev@proofable.me instead of opening a public issue. Leave keys, tokens, and private proof content out of the report.

What this package contains

@proofable/mcp publishes the registry manifest (server.json), the host plugins, and the public skills. Its proofable-mcp command forwards MCP requests to the hosted server and stores nothing. Proofable operates the hosted server at https://mcp.proofable.me/mcp; docs.proofable.me/security describes its security posture.

Authentication model

  • Interactive clients sign in with hosted OAuth (PKCE, silent refresh). Client config and chat never hold a key.
  • Servers, CI, and headless agents send a server key (npk_...) from PROOFABLE_ACCESS_KEY as a Bearer token. Rotate keys under Access keys.
  • Keep keys out of repositories, screenshots, chat, issue threads, and skill prompts.

Privacy defaults

The server returns the verification result and the proof record, never the private data used to complete a check. Proofs are private by default; publishing one takes an explicit visibility choice.

Signing and key custody

You generate and hold your private keys; Proofable never does. Dedicated agent accounts are created in your wallet or secure runtime, and only the public address reaches Proofable. A proof signature is bound to the request it signs and does not work as a bearer token.

Related

  • Wrong tool description: open an issue. The tool fields in server.json mirror the hosted server.
  • SDK signing and privacy defaults: proofable/sdk SECURITY.md.

There aren't any published security advisories