The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external web pages provided by the user, which is a vector for indirect prompt injection attacks.
Ingestion points: The skill uses context_dev_scrape_markdown and context_dev_extract_structured_data to import content from web sources into the agent's context (SKILL.md).
Boundary markers: The skill explicitly instructs the agent to "Ask before adding
don't auto-add without confirmation," providing a human-in-the-loop safety checkpoint (SKILL.md).
Capability inventory: The skill allows writing extracted content to a local TASKS.md file (SKILL.md).
Sanitization: There is no evidence of specific input sanitization or filtering for the content retrieved from external URLs before it is processed or stored.
[EXTERNAL_DOWNLOADS]: The skill utilizes tools to fetch content from user-supplied URLs.
Evidence: The skill instructions and the API map reference tools like context_dev_scrape_markdown and context_dev_extract_structured_data for importing data from web sources (references/sandbase-api-map.md).

task-management