Loading ToolBench
The notary of the x402 agent economy, by ForgeMesh Labs. Cryptographic receipts for AI outputs: notarize any model inference with a signed Ed25519 attestation, sha256 content hash, and Merkle chain-anchor on Base or Solana — $0.001 per call via x402 USDC micropayments. Verification is free and needs no wallet. We notarize the hash, never your prompts.
Strong tool definitions with complete schemas, clear descriptions, and proper tool annotations. All 6 tools have explicit input schemas with typed properties and descriptions. Tool names follow verb_noun convention (notarize_*, verify_*, get_*). Descriptions are detailed (100-250 chars) and explain cost, prerequisites, and privacy guarantees. However, output schemas are not documented, LLMs cannot predict response structure. Parameter descriptions are good but lack format constraints (e.g., ISO-8601 format for client_timestamp not enforced). Error handling guidance is absent, tools don't explain recovery paths for common failures (wallet missing, insufficient balance, invalid attestation_id).
FREE — fetch the public receipt for an attestation: content hash, model, timestamps, Ed25519 signature, and Merkle anchor proof. Raw prompt/response are never stored, so receipts contain proof material only. No wallet needed.
Notarize up to 20 AI inferences in one call — one signed attestation per record. Ideal for audit trails and agent pipelines. Costs $0.005 USDC via x402 (requires WALLET_PRIVATE_KEY for Base or SOLANA_PRIVATE_KEY for Solana).
Get a cryptographic receipt for one AI inference. Returns a signed Ed25519 attestation, sha256 content hash, and Merkle chain-anchor status for {prompt, response, model_id}. The notary does NOT store your prompt or response — only the hash is retained. Costs $0.001 USDC via x402 (requires WALLET_PRIVATE_KEY for Base or SOLANA_PRIVATE_KEY for Solana).
FREE — the notary's Ed25519 public key (base64, raw 32 bytes) for fully offline signature verification. No wallet needed.
FREE — live aggregate stats: total notarizations, 24h volume, top models by attestation count, sealed/anchored Merkle batches. No wallet needed.
Output schemas not documented. LLMs cannot predict response structure (e.g., what fields does get_receipt return? Is signature a string or object? What is the Merkle proof format?). This forces agents to guess and risks parsing errors.
No error handling guidance. Tools don't explain what to do if wallet is missing, balance insufficient, attestation_id invalid, or network fails. Agents have no recovery path.
client_timestamp parameter lacks format constraint. Description says 'ISO-8601' but no pattern or format field enforces it. LLMs may pass invalid timestamps (e.g., '2024-01-15 10:30' instead of '2024-01-15T10:30:00Z').
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 64 | 2026-07-28+ | v2 |
FREE — verify any attestation issued by the notary. Supply the attestation_id plus either the original content ({prompt, response, model_id}) or its content_hash. Returns the Ed25519 signature check, hash comparison, and a Merkle inclusion proof once the batch is sealed. No wallet needed.
verify_attestation has ambiguous parameter logic. Description says 'supply attestation_id plus either original content OR content_hash' but schema doesn't enforce mutual exclusivity. LLMs may pass both or neither, causing silent failures.
notarize_batch lacks per-item error reporting. If 1 of 20 records fails (e.g., invalid model_id), does the tool fail entirely or return partial results? Agents cannot distinguish success from partial failure.