mercury-autocategorize-transactions
Pass
Audited by Gen Agent Trust Hub on Jun 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes
mercuryCLI commands to fetch categories and transactions, and to update transaction records. These actions are fundamental to the skill's purpose and are mitigated by a mandatory user confirmation step. - [PROMPT_INJECTION]: There is a potential for indirect prompt injection because the agent processes counterparty names and transaction descriptions which are untrusted external inputs. Ingestion points: Mercury transaction data fetched in Step 2 of SKILL.md. Boundary markers: No specific delimiters or instructions are provided to the agent to ignore embedded commands in transaction data. Capability inventory: Transaction update capability via
mercury transactions updatein Step 6. Sanitization: The process includes a mandatory manual review table (Step 4) and user confirmation (Step 5) before any data is written.
Audit Metadata