The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill installs the firecrawl-cli package from the official NPM registry and GitHub repository (github.com/firecrawl/cli). These are official distribution channels for a well-known technology service.
[COMMAND_EXECUTION]: Uses the firecrawl and npx firecrawl commands to interact with the Firecrawl API. Access is scoped to these specific binaries via the allowed-tools configuration in the frontmatter.
[PROMPT_INJECTION]: The skill processes untrusted web data, creating a surface for indirect prompt injection. Ingestion points: Web content fetched via firecrawl search, scrape, and crawl commands (defined in SKILL.md). Boundary markers: Documentation in rules/security.md recommends using -o to isolate fetched content in a .firecrawl/ directory and utilizing incremental reads (e.g., grep, head) to limit context exposure. Capability inventory: The skill utilizes Bash(firecrawl *) for network fetching and file writing. Sanitization: rules/security.md explicitly instructs the agent to treat fetched data as untrusted third-party data and avoid following instructions found within the content.

firecrawl