The Agent Skills Directory

[DATA_EXPOSURE]: The skill reads several local files to gather context, including context/brand-style.md, context/content-calendar.md, context/best-performers.md, and the agent-specific configuration file .claude/product-marketing-context.md.
[EXTERNAL_DOWNLOADS]: It incorporates the Firecrawl MCP tool (mcp__firecrawl__firecrawl_scrape) to fetch content from external Threads profiles for tone and engagement analysis.
[COMMAND_EXECUTION]: The skill manages its own output by creating the outputs/threads/ directory and writing generated content files to the local file system.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests data from external web sources and local files that may contain instructions designed to influence the agent's behavior.
Ingestion points: External data is retrieved via Firecrawl scraping, and local data is read from the context/ directory.
Boundary markers: There are no instructions provided to wrap external or user-provided content in delimiters or to ignore embedded instructions within that data.
Capability inventory: The skill has the ability to read and write files, create directories, and perform network requests through configured MCP tools.
Sanitization: The skill does not define any methods for validating, filtering, or escaping content retrieved from external sources before it is added to the prompt context.

threads-writer