get_verification_artifact
Obtain a sparse verification artifact with raw calldata and decoding instructions for independent cross-check by a second LLM, enabling adversarial verification of transaction safety before Ledger approval.
Instructions
Return a sparse verification artifact for a prepared tx — raw calldata (or TRON rawDataHex), chain, to/value, payloadHash, preSignHash if preview_send has pinned gas, plus a static prompt instructing a second LLM on how to decode the bytes from scratch. Intended for adversarial independent verification: the user copies this artifact into a second LLM session (different provider recommended) so the second agent produces an independent decode with no shared context from the current conversation. If the two decodes disagree — or if the preSignHash doesn't match what Ledger displays at sign time — the user rejects. Does NOT call any external API; read-only in-memory lookup. Output deliberately omits the server's humanDecode, swiss-knife URL, and 4byte cross-check so the second agent cannot echo them. Handles live in-memory for 15 minutes after issue.
Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| handle | Yes | Opaque handle returned by any prepare_* tool. Returns a sparse, copy-paste-friendly JSON artifact carrying the raw calldata (or TRON rawDataHex), chain, recipient, value, payloadHash, and — when preview_send has already pinned gas — the Ledger blind-sign preSignHash. A static prompt telling a second LLM how to independently decode the bytes is included. The artifact intentionally omits the server's humanDecode, swiss-knife URL, and 4byte cross-check so the second agent cannot parrot them. |