tutorial-harvesting-engine

Pass

Audited by Gen Agent Trust Hub on Jun 22, 2026

Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [PROMPT_INJECTION]: The skill defines an attack surface for indirect prompt injection by processing external data sources.\n
  • Ingestion points: Untrusted data enters the agent context through the sources parameter (URLs, files, or directories) defined in SKILL.md.\n
  • Boundary markers: The instructions lack delimiters or explicit warnings to the agent to disregard instructions embedded within the harvested content.\n
  • Capability inventory: The skill is designed to find, parse, and extract content from documentation sites and file systems.\n
  • Sanitization: There is no evidence of filtering, escaping, or validation of the content extracted from external sources.\n- [DATA_EXFILTRATION]: The input schema allows the agent to access arbitrary 'files, or directories' on the local system. This represents a data exposure risk if the agent is prompted to harvest paths containing sensitive configuration or credentials.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 22, 2026, 02:30 AM
Security Audit — agent-trust-hub — tutorial-harvesting-engine