Skills
Navigate
skills/zapabob/skills/merger-model/Gen Agent Trust Hub

merger-model

Pass

Audited by Gen Agent Trust Hub on Sep 5, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external financial data which introduces a potential surface for indirect prompt injection.\n
  • Ingestion points: web_search, web_extract, and browser_navigate tools are used to retrieve information from public web pages and the SEC EDGAR database (SKILL.md).\n
  • Boundary markers: None explicitly defined for the ingested external data.\n
  • Capability inventory: The skill instructs the agent to execute a Python script (recalc.py) and perform file system operations using openpyxl (SKILL.md).\n
  • Sanitization: The skill mandates flagging unsourced data with a [UNSOURCED] label and strictly prohibits fabrication.\n- [COMMAND_EXECUTION]: The skill provides instructions to execute a local Python script (recalc.py) for workbook recalculation. This is a documented step within the financial modeling workflow.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 5, 2026, 12:08 PM