The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill manages a pipeline that ingests content from external websites (via Firecrawl) which is then processed by AI prompts for summarization and meaningfulness judging.
Ingestion points: External web pages scraped by the Firecrawl backend and delivered via webhooks as described in SKILL.md.
Boundary markers: No specific delimiters or safety warnings for the ingested content are mentioned in the skill instructions to prevent the model from obeying embedded instructions.
Capability inventory: The skill allows triggering administrative API workflows (sync, refetch-release, backfill-source) and modifying source metadata based on monitored content.
Sanitization: While a specific diff parser is used for technical extraction, there is no mention of sanitizing or filtering the natural language content of the ingested pages.
[COMMAND_EXECUTION]: The skill provides template curl commands for administrative tasks such as syncing monitors and backfilling data. These commands include headers for authentication via environment variables like $RELEASES_API_KEY and target a configurable $RELEASES_API_URL.

firecrawl-monitoring