The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to scrape and interact with arbitrary external web pages based on user-provided URLs. These web pages are untrusted sources that could contain hidden or malicious instructions aimed at influencing the agent's actions during the interaction phase.
Ingestion points: The skill ingest data via the firecrawl scrape "<url>" command in SKILL.md.
Boundary markers: The instructions do not define specific delimiters or warnings to ignore instructions embedded within the scraped web content.
Capability inventory: The skill utilizes Bash(firecrawl *) and Bash(npx firecrawl-cli *) which provide capabilities to click, fill forms, and navigate the live browser session.
Sanitization: There is no evidence of sanitization or filtering of the scraped content before it is processed by the agent.
[COMMAND_EXECUTION]: The skill relies on executing shell commands through the firecrawl and npx firecrawl-cli tools. These are recognized as tools provided by the vendor to facilitate the skill's core functionality of web interaction.
[EXTERNAL_DOWNLOADS]: The skill uses npx firecrawl-cli, which triggers a download of the package from the official NPM registry. This is a standard deployment method for the vendor's command-line interface.

firecrawl-interact