configure

Warn

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: MEDIUMDATA_EXFILTRATIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill configures a telemetry system designed to stream session events to an external webhook URL.
  • This includes streaming of all 18 event types, potentially exposing sensitive information such as user prompts (UserPromptSubmit), tool inputs (PreToolUse), and tool results (PostToolUse) to a third-party endpoint provided by the user.
  • [REMOTE_CODE_EXECUTION]: The configuration process enables several MCP servers that are downloaded and executed at runtime using npx -y from the npm registry.
  • Examples include @upstash/context7-mcp@latest, tavily-mcp@latest, and @21st-dev/magic@latest.
  • [EXTERNAL_DOWNLOADS]: The skill installs external dependencies such as the agentation and agentation-mcp packages using npm install during the integration setup.
  • [COMMAND_EXECUTION]: The skill modifies the user's shell profiles (~/.zshrc or ~/.bashrc) to persist environment variables like ENABLE_CLAUDEAI_MCP_SERVERS and ORCHESTKIT_HOOK_TOKEN.
  • [COMMAND_EXECUTION]: It executes generator scripts, such as npm run generate:http-hooks, to dynamically create and write configuration files to the user's filesystem based on interactive inputs.
Audit Metadata
Risk Level
MEDIUM
Analyzed
May 14, 2026, 10:18 PM
Security Audit — agent-trust-hub — configure