The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted titles and summaries from various external news sources and RSS feeds, which creates a surface for indirect prompt injection if those sources contain malicious instructions.\n
Ingestion points: Data is ingested via the fetch_hackernews, fetch_v2ex, and fetch_rss functions in scripts/scan.py.\n
Boundary markers: While the output report is structured using Markdown headers and bullet points, there are no explicit boundary markers or instructions to the agent to disregard embedded commands in the fetched data.\n
Capability inventory: The skill script scripts/scan.py performs network requests and local file writes, and the agent is instructed to perform web searches on platforms like X and Jike.\n
Sanitization: The retrieved news titles and summaries are not sanitized for potential injection patterns before being presented to the agent.\n- [EXTERNAL_DOWNLOADS]: The skill retrieves data from well-known industry news platforms and external APIs.\n
Fetches news signals from Hacker News using the Algolia API.\n
Accesses the V2EX API to retrieve trending technical topics.\n
Downloads content from user-configured RSS feeds such as 36kr and SSPAI.\n- [COMMAND_EXECUTION]: The skill workflow involves executing a local Python script scripts/scan.py to aggregate and filter signals, which is the primary mechanism for its core functionality.

industry-intelligence-radar