Intentrax
Intentrax issues signed proofs for consequential actions taken by AI agents, such as payments, infrastructure changes and regulated decisions. Anyone can check these proofs, without trusting the operator's logs. This connector brings that check into Claude. It has three read-only tools: - Verify a published proof. Look up a proof by its ID on Intentrax's public verifier, and get its status, verdict, and a per-dimension assessment of what was and was not verified. - Get the signing keys. Fetch Intentrax's published production or sandbox key set with its SHA-256, for checking proofs offline. - Verify a proof export. Paste an exported proof, and the connector runs the same verifier as `npx @intentrax/sdk verify-proof`, against the pinned production keys. Every answer states its limits. A proof records an Intentrax decision about one action. It does not by itself show that an external effect happened, or that every agent action was routed through Intentrax. The connector needs no account and no key. It never executes, admits, signs, or writes anything. It keeps nothing you send it: a pasted export is checked in a temporary folder that is deleted afterwards.
Details
https://mcp.intentrax.com/public · streamable-http
