The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and analyze content from external websites provided by the user or discovered via search. This exposes the agent to indirect prompt injection attacks where a third party could embed malicious instructions on a web page to hijack the agent's logic.
Ingestion points: External data is ingested via firecrawl_map, firecrawl_scrape, firecrawl_search, and WebFetch as described in SKILL.md (Research Process) and references/tool-reference.md.
Boundary markers: SKILL.md includes a "5. Untrusted Input" section that explicitly instructs the agent to treat fetched content as data only, never as instructions, and to ignore any attempts at AI-directed manipulation.
Capability inventory: The skill uses network-enabled MCP tools (Firecrawl, DataForSEO) and has the capability to write synthesized profiles and raw data to the local file system (competitor-profiles/ directory).
Sanitization: There is no automated sanitization of the scraped content; the skill relies on the agent's ability to follow instructions to ignore embedded directives within the raw markdown or text.

competitor-profiling