The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests untrusted data from external websites.
Ingestion points: Scrapes content from competitor URLs using firecrawl_scrape (SKILL.md).
Boundary markers: The instructions do not specify explicit delimiters or 'ignore' instructions for the scraped content during the synthesis phase.
Capability inventory: The agent has file-writing capabilities (to the competitor-profiles/ directory) and network access via MCP tools.
Sanitization: There is no mention of sanitizing or escaping the content retrieved from external sites before it is processed by the agent.
[DATA_EXPOSURE]: The skill attempts to read local project context files to gather information about the user's own product.
Evidence: It checks for the existence of .agents/product-marketing.md, .claude/product-marketing.md, or product-marketing-context.md (SKILL.md).
Context: While these are standard locations for agent-specific context in certain environments, users should be aware that the skill accesses these local files.

competitor-profiling