Use Runtime Copilot In Codex

This repository can be used in Codex as an MCP-connected operational surface.

The best way to think about it is:

What To Install

This repository now includes:

1. Connect The MCP Server

Configure Codex so that mini-data-engine points at the local MCP server entrypoint:

[mcp_servers.mini-data-engine]
command = "/Users/ll/Documents/data-engineering-runtime-lab/.venv/bin/python"
args = ["/Users/ll/Documents/data-engineering-runtime-lab/mcp_engine_server.py"]

If you prefer Docker, point Codex at the published MCP image instead.

2. Install The Skill

Copy the skill folder into your Codex skills directory:

mkdir -p "$HOME/.codex/skills/runtime-copilot"
cp -R /Users/ll/Documents/data-engineering-runtime-lab/codex/skills/runtime-copilot/. "$HOME/.codex/skills/runtime-copilot/"

After that, Codex can trigger the skill when the request is about runtime diagnostics, MCP discovery, explainability, or regression checks.

3. Use The Right Entry Points

For discovery:

For health and diagnostics:

For regression and release confidence:

4. Reuse The Automation Examples

The repo includes two ready-made automation examples:

They are meant as templates for:

5. What A Good Codex Session Looks Like

A good Codex session usually starts with:

  1. project_tool_catalog
  2. project_get_defaults
  3. one operational check such as health_check or project_run_regression

That keeps the interaction declarative and MCP-first instead of dropping straight into shell commands.