ad-angle-miner
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function involves ingesting and analyzing untrusted external data from multiple platforms, which represents a significant injection surface.
- Ingestion points: The skill scrapers collect data from G2, Capterra, Trustpilot, Reddit, Twitter/X, and user-provided support tickets or sales call transcripts.
- Boundary markers: The instructions do not define clear delimiters or "ignore instructions" guards for the AI when processing the scraped content in Phase 2.
- Capability inventory: The skill executes shell commands to run scrapers and writes analysis results to the filesystem (e.g.,
clients/<client-name>/ads/angle-bank-[YYYY-MM-DD].md). - Sanitization: There is no evidence of content sanitization or filtering to remove potential prompt injection payloads from the scraped text before analysis.
- [COMMAND_EXECUTION]: The skill instructions include several shell command patterns using
python3to run scraper scripts. These commands interpolate user-provided parameters such as product names, queries, and subreddit names directly into the command line, which could lead to command injection if the underlying scripts do not properly sanitize these arguments.
Audit Metadata