The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the bash tool to execute the exa-ai CLI and the jq utility for searching and processing web results.
[PROMPT_INJECTION]: The skill's primary function of fetching and summarizing external web content creates an indirect prompt injection surface where untrusted data could attempt to influence the agent's instructions. • Ingestion points: External web pages and summaries retrieved via exa-ai find-similar and exa-ai search in SKILL.md. • Boundary markers: The skill does not explicitly define delimiters or provide warnings to ignore embedded instructions in the fetched content. • Capability inventory: Execution of exa-ai and jq commands within a bash environment. • Sanitization: The skill recommends using JSON schemas (--summary-schema) and jq for structured extraction, which provides a layer of validation for the ingested data.

exa-find-similar