The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves scraping and analyzing large corpuses of untrusted data from the web, creating a potential surface for indirect prompt injection attacks.
Ingestion points: Phase 3 (Content Scraping) fetches data from Substack, Twitter, and dead domains via the Wayback Machine CDX API.
Boundary markers: Scraped content is stored in Markdown files formatted with YAML frontmatter headers to separate metadata from the content.
Capability inventory: The skill creates directories and writes files to the user's home directory (~/.claude/userModels/) and executes a local Python validation script.
Sanitization: The skill performs structural and verbatim quote validation using a Python script, but does not explicitly mention filtering or sanitizing the scraped content for malicious instructions before the analysis phase.
[COMMAND_EXECUTION]: The skill workflow includes a phase for automated validation that requires executing a local Python script (scripts/validate_usermodel.py) using the uv run command to check file structure and quote accuracy.
[EXTERNAL_DOWNLOADS]: The skill uses external services and tools to fetch remote data, specifically utilizing Firecrawl for Substack, Jina for Twitter content, and the Wayback Machine CDX API for archived domains.

user-model-builder