The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill fetches search results from Brave Search and returns them to the agent. This ingestion of untrusted external data could contain malicious instructions designed to hijack the agent's behavior.
Ingestion points: The brave-search script fetches live content from https://search.brave.com/search?q=$q.
Boundary markers: The script does not wrap the fetched content in specific boundary markers or tags to help the agent distinguish it from instructions.
Capability inventory: The skill itself primarily performs network reads and local file processing; risks depend on what the agent does with the output.
Sanitization: The script uses sd to remove some image tags and Brave-specific navigation markers, but does not perform safety filtering of the actual content text.
[COMMAND_EXECUTION]: The skill relies on and executes external CLI tools (scrapling and sd) to perform its search and text-processing functions.

brave-search