ad-angle-miner
Pass
Audited by Gen Agent Trust Hub on Apr 23, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local Python scripts located in the skills/ directory (e.g., scrape_reviews.py, scrape_reddit.py, scrape_twitter.py) to gather data from external platforms. These commands are integral to the skill's data collection phase.\n- [PROMPT_INJECTION]: The skill processes untrusted data from public reviews, Reddit threads, and social media posts. This creates a surface for indirect prompt injection. Ingestion points: Phase 1 data collection from external platforms and internal support tickets. Boundary markers: Absent. Capability inventory: The skill writes analysis results to the local filesystem and executes scripts. Sanitization: The instructions do not specify sanitization or validation of the scraped content before analysis, which is a common characteristic for this use case.
Audit Metadata