mails-so-automation

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS] (MEDIUM): The skill instructs the agent to connect to an external MCP endpoint https://rube.app/mcp. This source is not on the trusted repository or organization list, meaning the tools and schemas provided are controlled by an unverified third party.
  • [REMOTE_CODE_EXECUTION] (MEDIUM): The skill utilizes RUBE_MULTI_EXECUTE_TOOL and RUBE_REMOTE_WORKBENCH for operation. This effectively allows a remote server to define and execute logic on the agent's behalf, which can be risky if the third-party service is compromised or malicious.
  • [DATA_EXFILTRATION] (LOW): By design, the skill processes email data through the Rube MCP infrastructure. Users should be aware that sensitive email content will be transmitted to and processed by the rube.app services.
  • [PROMPT_INJECTION] (LOW): As a Category 8 (Indirect Prompt Injection) finding, the skill processes untrusted data (emails from Mails So). There is a risk that malicious emails could contain instructions designed to manipulate the agent's behavior via the available Rube tools.
  • Ingestion points: Email content fetched via mails_so toolkit tools (referenced in SKILL.md).
  • Boundary markers: None specified in the instructions to help the agent distinguish between system instructions and email content.
  • Capability inventory: RUBE_MULTI_EXECUTE_TOOL, RUBE_REMOTE_WORKBENCH, and RUBE_MANAGE_CONNECTIONS (SKILL.md).
  • Sanitization: No mention of sanitization or filtering for the processed email data.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 06:45 PM