tougu-writer-factory
Pass
Audited by Gen Agent Trust Hub on Apr 1, 2026
Risk Level: SAFEDATA_EXFILTRATIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill ingests user-provided articles during Phase 1 to perform style analysis and extraction. This represents an indirect prompt injection surface where malicious instructions embedded in sample text could attempt to subvert the agent's behavior during the analysis phase. 1. Ingestion points: Phase 1 article collection (references/phase1.md). 2. Boundary markers: Absent. 3. Capability inventory: Local file writes and network data fetching via Python scripts. 4. Sanitization: Absent.
- [DATA_EXFILTRATION]: The market_data.py script and associated documentation reference the use of the akshare library and the requests module to fetch market data from external financial providers such as East Money and Tencent. These outbound network requests are legitimate and necessary for the skill's data aggregation functions.
- [EXTERNAL_DOWNLOADS]: The assets/html_template.html file retrieves styling and font resources from well-known services, specifically Tailwind CSS (cdn.tailwindcss.com) and Google Fonts (fonts.googleapis.com), to facilitate the rendering of visual posters.
- [COMMAND_EXECUTION]: The skill includes Python scripts (market_data.py, self_eval.py) that perform file system operations to organize generated skills in the output directory and execute data processing logic. These scripts function within the expected operational scope of a writing factory skill.
Audit Metadata