The Agent Skills Directory

[SAFE]: The skill is primarily instructional and does not include any executable code or scripts. It defines workflows for research, CRM management, and opportunity analysis using a remote MCP server.
[SAFE]: Proactive safety measures are integrated into the instructions to prevent indirect prompt injection. Specifically, the skill directs the agent to treat retrieved documents as evidence only and to ignore any instructions found within them that attempt to alter workflows or expose data.
[SAFE]: The skill emphasizes credential security by instructing the agent to never request keys in chat, place them in links, or send them to third-party destinations. It explicitly directs the use of the client's connection settings for authentication.

govcon-mcp