The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes content from external URLs. The ingested markdown is provided to the agent without boundary markers or warnings to ignore instructions found within the scraped content. 1. Ingestion points: Data enters the agent context via firecrawl scrape (SKILL.md). 2. Boundary markers: Absent. 3. Capability inventory: Uses the Bash tool. 4. Sanitization: Absent.
[COMMAND_EXECUTION]: The skill executes shell commands, including firecrawl and npx firecrawl, to interact with the web and local file system.
[EXTERNAL_DOWNLOADS]: The skill uses npx firecrawl, which downloads and executes the firecrawl package from the npm registry if not already present.

firecrawl-scrape