rudder-setup

Pass

Audited by Gen Agent Trust Hub on Nov 5, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: Downloads and executes an installation script for rudder-cli from the official RudderLabs GitHub repository. This download is necessary for the skill's operation and originates from the vendor's own infrastructure.
  • [COMMAND_EXECUTION]: Utilizes bash and shell commands for environment discovery, tool verification, and executing included scripts for event validation and run recording.
  • [EXTERNAL_DOWNLOADS]: Installs system dependencies like jq and platform-specific RudderStack SDKs to facilitate the instrumentation of the user's application.
  • [PRIVILEGE_ESCALATION]: Uses sudo commands specifically to install the jq dependency on Linux systems when it is not already available.
  • [INDIRECT_PROMPT_INJECTION]: Processes external data from MCP servers and webhooks. The skill defines strict boundaries, instructing the agent to treat these inputs as data and disregard any embedded commands.
  • [DYNAMIC_EXECUTION]: Executes local JQ and shell scripts to perform environment probing, stateful undo management, and tracking validation.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 5, 2026, 05:49 PM