The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill enables agents to retrieve and process external email content through tools like keyid_get_inbox and keyid_get_thread. This creates a vulnerability surface where malicious instructions embedded in incoming emails could influence the agent's actions.
Ingestion points: External email data ingested via the keyid_get_inbox tool and webhook integrations described in SKILL.md.
Boundary markers: The instructions lack specific delimiters or instructions to the agent to treat email content as untrusted data.
Capability inventory: The skill provides significant capabilities, including sending emails (keyid_send), managing contacts, and configuring auto-replies or webhooks.
Sanitization: No sanitization or filtering logic for incoming email bodies is mentioned in the skill documentation.
[EXTERNAL_DOWNLOADS]: The skill requires the installation or execution of the @keyid/agent-kit package from the NPM registry and communicates with the keyid.ai API for its core functionality.
[COMMAND_EXECUTION]: Configuration for Claude Desktop and Cursor involves executing shell commands via npx to run the MCP server, which is a standard procedure for this protocol.

keyid-agent-kit-mcp