lov-bp-polish
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data (Business Plan outlines in Markdown, PDF, and PPTX formats) to provide feedback. This creates a surface where malicious instructions embedded in the input documents could attempt to influence the agent's review or subsequent actions.\n
- Ingestion points: Processes
./business-plan/outline.mdand user-provided files during the audit workflow.\n - Boundary markers: No explicit delimiters or 'ignore' instructions are used to isolate input data from the agent's core instructions.\n
- Capability inventory: Executes a local Python linter and performs file read/write operations within the workspace.\n
- Sanitization: No sanitization or escaping of input text is performed before it is processed by the agent.\n- [COMMAND_EXECUTION]: The skill executes a bundled Python script as part of its deterministic audit workflow.\n
- Evidence:
python3 "$SKILL_DIR/scripts/audit_bp.py"is called inSKILL.mdto analyze the outline and generate a markdown report. The script uses standard libraries and performs regex-based parsing of the input text.
Audit Metadata