Install n8n-mcp and skills to OpenAI Codex

[sanchomuzax]

I've solved it so that others don't have to struggle with it anymore. I have successfully installed both the mcp-server and skill components on a Raspberry Pi Docker-based solution, alongside VSCode running OpenAI Codex.

N8N MCP SERVER - REPRODUCIBLE SETUP GUIDE (RASPBERRY PI 5)
Version: 1.15

1. DOCKER SETUP (CONTAINERIZATION)

---

mkdir -p ~/mcp-servers/n8n
cd ~/mcp-servers/n8n
nano Dockerfile

[ Dockerfile content ]
FROM node:20-slim
WORKDIR /app
RUN npm config set fetch-retries 5 && \
npm config set fetch-retry-mintimeout 20000 && \
npm config set fetch-retry-maxtimeout 120000 && \
npm install -g n8n-mcp --legacy-peer-deps
ENTRYPOINT ["n8n-mcp"]
[ Dockerfile end ]

docker build -t n8n-mcp:1.15 .
docker tag n8n-mcp:1.15 n8n-mcp:latest

2. CLIENT CONFIGURATION (config.toml)

---

Path: ~/.config/codex/config.toml

[mcp_servers.n8n-mcp]
command = "docker"
args = [
"run",
"-i",
"--rm",
"-e", "N8N_API_KEY=YOUR_N8N_API_KEY",
"-e", "N8N_BASE_URL=http://172.17.0.1:5678",
"n8n-mcp:latest"
]
startup_timeout_sec = 25
tool_timeout_sec = 60

3. SKILL CREATOR INSTRUCTION

---

Skill creator prompt:

[$skill-creator](~/.codex/skills/.system/skill-creator/SKILL.md) Use the n8n MCP server. Learn the structure of n8n v2 nodes from the [https://github.com/czlonkowski/n8n-skills/tree/main/skills](https://github.com/czlonkowski/n8n-skills/tree/main/skills) source. Your task: designing n8n workflows, generating JSON structures, and debugging based on execution logs.

4. DIAGNOSTICS

---

• Check tools: list_mcp_tools command.
• Restart: Mindig indítsd újra a Codex session-t a config módosítása után.
• Manual test:
  docker run -i --rm -e N8N_API_KEY=... -e N8N_BASE_URL=... n8n-mcp:latest
Input: {"jsonrpc":"2.0","method":"list_tools","id":1}
  ==========================================================