Skills
skills/aradotso/mcp-skills/codex-control-plane-mcp/Gen Agent Trust Hub

codex-control-plane-mcp

Pass

Audited by Gen Agent Trust Hub on Jun 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides several command-line utilities for administration and environment management, such as codex-control-plane-mcp-admin for initialization and codex-control-plane-mcp-hooks for installing and managing system hooks for Codex Desktop.
  • [EXTERNAL_DOWNLOADS]: The documentation includes instructions for installing the primary package from the vendor's GitHub repository (github.com/aresyn/codex-control-plane-mcp.git) and public package registries using tools like pipx and uvx.
  • [PROMPT_INJECTION]: The skill processes external data that could contain malicious instructions (indirect prompt injection surface).
  • Ingestion points: Untrusted data enters the agent context through the prompt parameter in codex_submit_task and the feedback parameter in codex_approve_plan, as well as through chat history read from codex_get_chat.
  • Boundary markers: No explicit delimiter or instruction-guardrail markers are defined for the prompt interpolation.
  • Capability inventory: The skill possesses capabilities to orchestrate Codex Desktop tasks, manage local SQLite databases, install system hooks, and perform automated repairs via codex_repair_issue.
  • Sanitization: There is no mention of sanitization or filtering for the data interpolated into the task prompts.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 17, 2026, 08:46 PM
Security Audit — agent-trust-hub — codex-control-plane-mcp