The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill connects to a hosted Model Context Protocol (MCP) server at https://warmysender.com/mcp and references the @warmysender/mcp package. These are vendor-owned resources used to provide the tools for outreach and lead management.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources such as lead database results and prospect lists. Ingestion points: Data from search results and prospect imports are ingested into the agent context via tools described in SKILL.md. Boundary markers: The skill includes explicit 'Hard rules' (Section 2) instructing the agent to report facts, avoid guessing, and seek user confirmation for all destructive actions. Capability inventory: The agent interacts with remote tools for outreach scheduling; it lacks local filesystem write or code execution capabilities. Sanitization: The architecture relies on an external server-side scheduler that enforces safety limits and pacing to prevent the agent's actions from causing harm, effectively mitigating injection risks.

warmysender