The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill retrieves content from external websites via the Brave Search API, which introduces a surface for indirect prompt injection where malicious instructions could be embedded in search results.
Ingestion points: brave_search.py in the web_search and local_search functions which fetch third-party web snippets.
Boundary markers: Absent; the skill returns raw result strings without additional delimiters or instruction-bypass warnings.
Capability inventory: The skill scripts do not contain dangerous capabilities such as arbitrary subprocess execution, file writing, or system modification based on the ingested data.
Sanitization: Absent; web content (titles and descriptions) is passed through to the agent without filtering or escaping.

brave-search