The Agent Skills Directory

[EXTERNAL_DOWNLOADS] (MEDIUM): The skill requires adding an external MCP server from https://rube.app/mcp. This domain is not included in the 'Trusted External Sources' list, necessitating manual review of the provider.\n- [REMOTE_CODE_EXECUTION] (MEDIUM): The use of RUBE_MULTI_EXECUTE_TOOL and RUBE_REMOTE_WORKBENCH allows the remote server to dictate tool execution logic within the agent context, effectively functioning as a remote orchestration layer.\n- [PROMPT_INJECTION] (LOW): The skill processes data from Missive (emails), creating an attack surface for indirect prompt injection. Mandatory Evidence Chain: 1. Ingestion points: Email content and thread metadata. 2. Boundary markers: None present in the instructions. 3. Capability inventory: Significant capability to modify/send data via RUBE_MULTI_EXECUTE_TOOL. 4. Sanitization: No sanitization or safety delimiters are described.\n- [NO_CODE] (SAFE): The skill consists entirely of configuration and instructions in markdown, with no local script files provided.

missive-automation