The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted local files (PDF, DOCX, etc.) which may contain hidden instructions that could influence the agent's behavior when it subsequently reads the parsed markdown content.
Ingestion points: The firecrawl parse command ingest arbitrary user-specified local files.
Boundary markers: The skill does not define delimiters or provide warnings to help the agent distinguish between document content and system instructions.
Capability inventory: The agent is instructed to use shell commands like head, grep, and rg to interact with the potentially malicious parsed data.
Sanitization: No sanitization or validation of the extracted text is mentioned in the skill instructions.
[DATA_EXFILTRATION]: Local document contents are sent to the external Firecrawl service for parsing, which constitutes an outbound transfer of local data to a third-party endpoint.
[COMMAND_EXECUTION]: The skill relies on the execution of the firecrawl CLI tool and standard Unix utilities to manage local directories and read file contents.

firecrawl-parse