The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes the firecrawl command-line interface or npx firecrawl-cli to perform web scraping operations. These commands are explicitly permitted in the allowed-tools configuration. The skill also suggests using standard shell utilities like head and grep to inspect the results.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted content from external URLs.
Ingestion points: Data enters the system via firecrawl scrape "<url>", which fetches content from any user-provided or agent-discovered website.
Boundary markers: The instructions do not define strict boundary markers for the scraped content, though it is converted to markdown to be more legible for the agent.
Capability inventory: The agent has access to local shell execution (firecrawl, npx, head, grep) and file system writes (via the -o flag to the .firecrawl/ directory).
Sanitization: The Firecrawl service performs sanitization by converting raw HTML and JavaScript-rendered content into clean markdown, which reduces the surface area for direct script execution but does not prevent natural language instructions from appearing in the output.

firecrawl-scrape