The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions provide various examples of using the firecrawl CLI and npx firecrawl-cli to extract content from web pages and save it to the local filesystem.
[EXTERNAL_DOWNLOADS]: The skill utilizes npx to fetch and run the firecrawl-cli tool. The package is a vendor-owned resource used to provide the skill's primary functionality.
[INDIRECT_PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by fetching content from arbitrary external URLs and providing the resulting markdown to the agent. This is an inherent risk of web scraping tools.
- Ingestion points: Untrusted content is ingested from external URLs via the firecrawl scrape command as described in SKILL.md.
- Boundary markers: Absent; there are no specific instructions provided to the agent to treat the scraped content as data rather than instructions.
- Capability inventory: The skill can execute the firecrawl command-line tool, which includes network access and the ability to write files to the .firecrawl/ directory.
- Sanitization: The tool filters navigation and footers when requested and converts HTML to markdown, which provides basic structural sanitization.

firecrawl-scrape