rudder-mcp-workflow

Pass

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches and executes the mcp-remote utility from the public npm registry to facilitate a connection to the RudderStack hosted MCP server.
  • [COMMAND_EXECUTION]: Provides a configuration snippet that executes a command to initialize the remote MCP transport layer.
  • [PROMPT_INJECTION]: Defines a surface for indirect prompt injection as the agent interacts with external data (RudderStack workspace resources and live events). The skill proactively provides defensive guidelines for processing this data.
  • Ingestion points: Data retrieved via MCP tools from RudderStack's hosted server (SKILL.md).
  • Boundary markers: Advises the use of defined schemas and ignoring unexpected keys to ensure data integrity (SKILL.md).
  • Capability inventory: Capability to connect to and manage RudderStack resources via the mcp-remote client configuration (SKILL.md).
  • Sanitization: Directs the agent to validate IDs and names while sanitizing results from warehouse queries to prevent injection from destination data (SKILL.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 16, 2026, 05:30 AM
Security Audit — agent-trust-hub — rudder-mcp-workflow