The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill documentation suggests using npx firecrawl-cli, which downloads the official Firecrawl package from the npm registry. This is a common method for accessing the service's features.
[COMMAND_EXECUTION]: The skill uses the firecrawl CLI to perform operations such as creating, updating, and running monitors. These commands are legitimate uses of the tool to interact with its cloud API.
[DYNAMIC_EXECUTION]: In structured tracking mode, the skill allows defining extraction logic through a prompt and schema passed via JSON. This enables dynamic data processing by the underlying scraping engine based on user-provided templates.
[INDIRECT_PROMPT_INJECTION]: The skill monitors external web content and triggers alerts based on an AI-driven goal. This creates a surface where malicious content on a monitored site could potentially influence the alert logic.
Ingestion points: Web content is ingested from external URLs specified in SKILL.md (e.g., via the --page or --crawl-url flags).
Boundary markers: There are no explicit instructions or delimiters mentioned to separate the scraped data from the AI judging instructions.
Capability inventory: The skill can retrieve web data and send notifications via webhooks and email, as documented in SKILL.md.
Sanitization: No explicit sanitization of the scraped website content is described before it is evaluated by the AI change judge.

firecrawl-monitor