future-compounder

Pass

Audited by Gen Agent Trust Hub on Aug 23, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes external data from financial filings, investor materials, and earnings calls, which represents a surface for indirect prompt injection.\n
  • Ingestion points: Primary filings and qualitative investor materials are read in the market-growth-intelligence and business-economic-engine stages (SKILL.md).\n
  • Boundary markers: No specific delimiters or instructions to ignore embedded commands are used when processing external text.\n
  • Capability inventory: The skill writes JSON data packs to the local filesystem and executes a provided Python validation script (SKILL.md, scripts/validate_pack.py).\n
  • Sanitization: The scripts/validate_pack.py utility performs schema and logic validation but does not filter for potential prompt injection content.\n- [COMMAND_EXECUTION]: The skill orchestrates analysis by executing a local Python script to validate data handoffs.\n
  • Evidence: The Operating rules section in SKILL.md specifies the execution of python scripts/validate_pack.py to ensure data integrity between analysis layers.\n
  • Assessment: The script is bundled with the skill, uses only standard libraries, and performs benign data validation tasks.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 23, 2026, 01:32 PM
Security Audit — agent-trust-hub — future-compounder