The Agent Skills Directory

[PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill fetches content from arbitrary external websites and processes it via the Firecrawl API, returning the results to the agent's context. This presents a surface for malicious instructions embedded in web content to influence the agent's behavior.
Ingestion points: Functions scrape, crawl, and search in scripts/firecrawl.sh retrieve external data.
Boundary markers: Absent. The instructions do not specify any delimiters or warnings to the agent to disregard instructions within the scraped data.
Capability inventory: The skill has access to the Bash tool, Read, Write, and Edit permissions, and performs network operations via curl.
Sanitization: The script uses jq to parse API responses, but the actual content (markdown or text) is returned to the agent without sanitization.
[COMMAND_EXECUTION]: The skill implements its logic through a Bash script (scripts/firecrawl.sh) which is executed with user-provided arguments such as URLs and search queries.
[CREDENTIALS_UNSAFE]: The script programmatically accesses a sensitive API key using the local pass password manager (pass show api/firecrawl). While this is a common method for CLI tools to manage secrets, it allows the agent to retrieve managed credentials.

firecrawl