The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from external websites via scraping and parsing techniques.
Ingestion points: Untrusted data enters the agent context through curl commands and HTML parsing logic (e.g., BeautifulSoup in Python snippets) used to fetch and process remote web pages (SKILL.md).
Boundary markers: The instructions focus on structured data (JSON-LD, Meta tags) but do not explicitly define boundary markers or 'ignore' instructions for the LLM when processing this external content.
Capability inventory: The skill uses curl for network requests and provides Python logic for DOM traversal and field extraction.
Sanitization: The skill includes a clean(s) function that performs HTML unescaping and Unicode normalization (NFKC) to handle non-breaking spaces and odd quotes.
[EXTERNAL_DOWNLOADS]: The skill references external resources and configuration commands.
Vendor Tools: It provides a command to register an external Model Context Protocol (MCP) server: claude mcp add --transport http manypi https://mcp.manypi.com/mcp -s user (SKILL.md).
External Repositories: It references a GitHub repository manypicom/sales-skills for related functionalities like lead data scraping and CRM hygiene (SKILL.md).
Network Requests: The provided examples use curl to fetch content from external URLs (e.g., example.com, manypi.com).

html-extraction-patterns