The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from arbitrary competitor websites and review platforms (such as G2 and Capterra) using tools like firecrawl_scrape and firecrawl_search. This external content is then processed to generate synthesized profiles, creating a surface for indirect prompt injection.
Ingestion points: Untrusted data enters the agent context via site mapping and scraping operations in Phase 1 (Firecrawl) and Phase 2 (DataForSEO).
Boundary markers: The instructions do not specify the use of clear delimiters or 'ignore embedded instructions' warnings when the agent processes the scraped raw markdown or JSON data during the synthesis phase.
Capability inventory: The agent has the capability to write multiple files to the local project directory (competitor-profiles/) based on its analysis of the ingested content.
Sanitization: While the skill provides instructions to remove cookies, tokens, and personal data from raw captures, it does not include measures to sanitize or escape potentially malicious instructions hidden in the scraped website content.
[EXTERNAL_DOWNLOADS]: The skill is designed to fetch information from external competitor URLs and SEO service providers to fulfill its primary research purpose.
Evidence: Phase 1 and Phase 2 describe processes for retrieving data from user-provided URLs and third-party APIs (Firecrawl, DataForSEO) for analysis and local storage.

competitor-profiling