The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the retrieval of untrusted content from the web via firecrawl search and the --scrape flag. This external data is ingested into the agent's context, which could allow instructions embedded in web pages to influence the agent's behavior.
Ingestion points: Web search results and scraped page content defined in SKILL.md.
Boundary markers: Absent. There are no instructions provided to separate untrusted web content or to ignore potential instructions within it.
Capability inventory: The skill uses Bash (via allowed-tools) to execute CLI commands and process results with jq.
Sanitization: Absent. The skill instructions suggest writing raw data to JSON files in the .firecrawl/ directory for subsequent agent processing without explicit filtering.
[COMMAND_EXECUTION]: The skill utilizes Bash to run firecrawl and npx firecrawl-cli for its core operations, including searching and sending feedback. These are vendor-provided tools intended for the skill's functionality.

firecrawl-search