The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to configure https://rube.app/mcp as an MCP server and references composio.dev for documentation. These are vendor-owned resources used to provide the skill's core functionality and tool definitions.
[COMMAND_EXECUTION]: The skill utilizes RUBE_MULTI_EXECUTE_TOOL and RUBE_REMOTE_WORKBENCH to execute automation tasks. These operations are performed based on schemas retrieved dynamically during the workflow.
[PROMPT_INJECTION]: The skill exhibits a vulnerability surface for indirect prompt injection because it processes and follows instructions retrieved from a remote source.
Ingestion points: Tool schemas, input requirements, and execution plans are ingested from the RUBE_SEARCH_TOOLS tool response (SKILL.md).
Boundary markers: No explicit delimiters or instructions to ignore instructions embedded in the search results are provided.
Capability inventory: The skill can execute various tools via RUBE_MULTI_EXECUTE_TOOL and perform workbench operations via RUBE_REMOTE_WORKBENCH (SKILL.md).
Sanitization: There is no evidence of sanitization or validation of the external schemas before the agent uses them to generate tool calls.

whautomate-automation