capture
Pass
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user brain dumps to automate workspace searches and propose file generation, creating an attack surface for indirect instructions.\n
- Ingestion points: User text dumps processed via triggers in SKILL.md.\n
- Boundary markers: Includes an 'Approval Gate' requiring explicit user consent before any action beyond organization is taken.\n
- Capability inventory: Uses scripts/workspace_inventory.py for filesystem scanning and offers file creation (e.g., plan drafting) in the Section 4 offers.\n
- Sanitization: Employs re.escape() in the inventory script to prevent regex injection from user-derived keywords.\n- [COMMAND_EXECUTION]: The skill executes internal Python scripts to perform core processing tasks.\n
- Invokes scripts/workspace_inventory.py, scripts/dump_classifier.py, and scripts/complexity_estimator.py to analyze dumps.\n
- All provided scripts use only Python standard libraries and do not perform network operations.
Audit Metadata