The Agent Skills Directory

[COMMAND_EXECUTION] (LOW): The script scripts/convert_academic.py uses subprocess.run to call external binaries (pandoc, myst). Although it uses the safer list-based argument passing (avoiding shell=True), it still executes external system processes based on input paths provided to the script.
[EXTERNAL_DOWNLOADS] (LOW): The script scripts/firecrawl_research.py communicates with the FireCrawl API (api.firecrawl.dev) to perform web searches and content scraping. This is the intended primary behavior of the skill.
[INDIRECT_PROMPT_INJECTION] (LOW): The skill ingests untrusted data from the web (scraped content) and writes it into markdown files without sanitization or boundary markers. This creates a surface where malicious instructions on a web page could influence the agent's behavior when it later reads the research notes.
Ingestion points: scripts/firecrawl_research.py (via FireCrawl API search results)
Boundary markers: Absent; the script writes raw markdown content retrieved from the API into local files.
Capability inventory: Includes subprocess execution of document converters and file write operations.
Sanitization: Absent; the script truncates content length for readability but does not filter for prompt injection patterns.

firecrawl-research