# Trust Standard Protocol > An open protocol that makes AI decision evidence durable, tamper-evident, and verifiable offline — by anyone, without trusting the vendor. It signs a canonical record of an AI-mediated event with Ed25519, hashes it with SHA-256 over canonical (JCS-style) JSON, and chains it to the previous record. TSP reports evidence integrity. It does not make you compliant, prove a model is correct, or confer official status. ## Core surfaces - [Verify a receipt](https://truststandardprotocol.com/verify/): Re-run TSP verification in the browser against the issuer public key. Nothing is uploaded; one changed byte fails. - [Playground](https://truststandardprotocol.com/playground/): Build a real chain of signed receipts, verify it, and break one byte to see verification fail live. - [Protocol](https://truststandardprotocol.com/protocol/): TrustEnvelope, TrustManifest, verification semantics, conformance, governance. ## Reference - [Tools](https://truststandardprotocol.com/tools/): Verifier cores (one signer, N verifiers), open MCP server/proxy, TrustBadge, operator modules. - [Why](https://truststandardprotocol.com/why/): The live-now pressure to keep AI evidence — EU AI Act (Art 12 / 14 / 19 / 50), DORA, US state laws, insurance — as evidence, not a verdict. - [Leaderboard](https://truststandardprotocol.com/leaderboard/): Public trust ranking recomputed from an org's own receipts. - [Hard questions](https://truststandardprotocol.com/hard-questions/): Privacy, forgery, governance — answered with the threat model in the open. - [Pricing](https://truststandardprotocol.com/pricing/): Free to verify, priced to operate. Official status is a governed grant, never a tier. - [Docs](https://truststandardprotocol.com/docs/): Quickstart (full reference site in Phase A.2). ## Seal (paid layer) - [Seal](https://truststandardprotocol.com/seal/): The independent notary — anchoring (Merkle root to the Evidence-ledger), qualified eIDAS / RFC 3161 time, and the register. You cannot notarize your own document; the free local layer is self-attested, the paid Seal is independently attested. Does not make you compliant. - [Conformance](https://truststandardprotocol.com/conformance/): The TSP-Certified mark — open standard, governed mark. Conformance is verified against the register, never self-declared; payment never grants official status. ## Company & governance - [Trust](https://truststandardprotocol.com/trust/): Who stands behind it, neutrality by determinism, credible-commitment bands, and the eIDAS / QTSP status (we are not yet a QTSP). - [Register](https://truststandardprotocol.com/register/): The public transparency-log backstop — don't trust us, verify us. Canonical-primary plus a public backstop so evidence survives us. - [Status](https://truststandardprotocol.com/status/): Operational continuity (static placeholder; no live infrastructure yet). - [Security](https://truststandardprotocol.com/security/): Coordinated disclosure and security posture. ## Machine-readable - [/.well-known/tsp.json](https://truststandardprotocol.com/.well-known/tsp.json): Protocol surface descriptor. - [/.well-known/security.txt](https://truststandardprotocol.com/.well-known/security.txt): Coordinated disclosure contact. - [/agent/tsp-manifest.json](https://truststandardprotocol.com/agent/tsp-manifest.json): Demo issuer manifest (public key). - [/agent/openapi.json](https://truststandardprotocol.com/agent/openapi.json): Verify API descriptor (placeholder). - [/agent/protocol.md](https://truststandardprotocol.com/agent/protocol.md): Protocol summary for agents. ## Boundary TSP verifies evidence state — integrity, provenance, custody, signature, manifest. It does not decide truth, correctness, legality, safety, or compliance. A person still decides whether to trust, escalate, audit, or reject. Packages use the @trust-standard-protocol/* scope but are not yet on npm — build from source at https://github.com/trust-standard-protocol/tsp; the open layer is Apache-2.0, the closed seal under a commercial TSP license.