last30days

Pass

Audited by Gen Agent Trust Hub on Mar 15, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface due to its core function of ingesting and processing untrusted data from social media and the web.
  • Ingestion points: External data from Reddit threads, X posts, and various web pages is loaded into the agent's context through search results processed in scripts/lib/openai_reddit.py, scripts/lib/xai_x.py, and scripts/lib/websearch.py.
  • Boundary markers: There are no explicit delimiters or strict "ignore embedded instructions" warnings provided for the research content before it is processed by the "Judge Agent" instructions in SKILL.md.
  • Capability inventory: The skill is capable of executing local Python scripts, making network requests via urllib.request, and writing output files to ~/.local/share/last30days/out.
  • Sanitization: While the skill performs date-based filtering and normalization, it does not sanitize the text content of fetched posts or pages for potential instruction injection.
  • [EXTERNAL_DOWNLOADS]: The skill performs network requests to well-known and trusted services, including the OpenAI and xAI APIs for discovery and Reddit for thread enrichment.
  • [COMMAND_EXECUTION]: Executes the skill's own internal logic via python3 ./scripts/last30days.py to handle data normalization, scoring, and output generation.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 15, 2026, 05:19 PM