The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill utilizes WebSearch and WebFetch tools to gather external macro-environmental evidence, creating an ingestion surface for potential indirect prompt injection embedded in retrieved web content.\n- Ingestion points: The workflow involves fetching external information across political, economic, social, technological, environmental, and legal factors from various web sources as described in the frontmatter and evidence plan.\n- Boundary markers: The skill emphasizes qualitative checks such as evidence quality, source dates, and bias detection, though it does not define strict technical or structural delimiters for the ingested data.\n- Capability inventory: The skill's capabilities are restricted to the allowed tools: Read, Grep, WebSearch, and WebFetch. It does not include tools for arbitrary command execution or system-level modification.\n- Sanitization: No automated sanitization of fetched content is specified; the framework relies on manual verification of source authenticity and recency by the agent.

pestel