The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the addition of a remote MCP server at https://rube.app/mcp. This external source provides the tool definitions and execution logic for the DocuSign automation, creating a dependency on unverified third-party infrastructure.
[REMOTE_CODE_EXECUTION]: By configuring an external MCP server, the agent dynamically retrieves and executes tool logic from a remote domain. If the provider (rube.app) is compromised or malicious, they could alter tool behavior to perform unauthorized actions on behalf of the user.
[DATA_EXFILTRATION]: The skill instructs users to use RUBE_MANAGE_CONNECTIONS for DocuSign OAuth. This process routes sensitive authentication flows and potentially document metadata through the third-party rube.app proxy, posing a risk of session token harvesting or unauthorized access to DocuSign account data.
[PROMPT_INJECTION]: The skill contains an indirect prompt injection surface. It reads external data from DocuSign templates and envelopes (using DOCUSIGN_GET_TEMPLATE and DOCUSIGN_GET_ENVELOPE) and has the capability to send documents for signature. Evidence chain:
Ingestion points: DocuSign template descriptions and recipient fields retrieved via API (SKILL.md).
Boundary markers: No specific delimiters or safety instructions are provided to handle malicious content inside templates.
Capability inventory: Has the ability to send emails and manage envelopes via DOCUSIGN_SEND_ENVELOPE (SKILL.md).
Sanitization: No evidence of input validation or sanitization for data retrieved from the DocuSign API before it is processed by the agent.

docusign-automation