skills/hazat/pi-config/add-mcp-server/Gen Agent Trust Hub

add-mcp-server

Pass

Audited by Gen Agent Trust Hub on May 3, 2026

Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill guides the agent to define and write shell commands into the mcp.json configuration file, which the system uses to launch local processes.
  • [REMOTE_CODE_EXECUTION]: It facilitates the setup of servers that can dynamically download and run external code (e.g., using npx) or connect to remote HTTP endpoints.
  • [DATA_EXFILTRATION]: The configuration steps involve collecting and persisting sensitive information such as bearer tokens, headers, and environment variables in local configuration files, and defining external URLs for data transfer.
  • [PROMPT_INJECTION]: The skill defines a surface for indirect prompt injection as it ingests and merges configuration data from user prompts or files (ingestion point) into mcp.json without boundary markers or sanitization. This is coupled with the capability to execute arbitrary commands or network requests defined within those configurations.
Audit Metadata
Risk Level
SAFE
Analyzed
May 3, 2026, 11:03 AM