Guides
Codex, Claude, and MCP Installation Guide
Fast path: run the install command for your agent, then verify with `/mcp`.
Official References
1. One-Liner Installer
Ready-to-run for your repo:
Universal command
curl -fsSL https://hard-wired.org/install-mcp.sh | bash -s -- \ --repo https://github.com/tappress/hard-wired-mcp-server.git \ --name hard-wired-mcp
2. Codex One-Liner (Auto Config)
If you know your Codex MCP config file path, set `--config` and it writes JSON for you.
Codex command
curl -fsSL https://hard-wired.org/install-mcp.sh | bash -s -- \ --repo https://github.com/tappress/hard-wired-mcp-server.git \ --name hard-wired-mcp \ --target codex \ --config /ABSOLUTE/PATH/TO/CODEX_MCP_CONFIG.json
- Run command.
- Restart Codex.
3. Claude Marketplace Install (Recommended)
Add the Hard Wired marketplace, then install the plugin. MCP config loads automatically from the plugin's `.mcp.json`.
Claude commands
/plugin marketplace add tappress/hard-wired-mcp-server /plugin install hard-wired@hard-wired-plugins
Plugin MCP server source: https://github.com/tappress/hard-wired-mcp-server.git
- Run the two `/plugin` commands in Claude Code.
- Run `/mcp` and confirm server `hard-wired` is loaded.
- Call `hardwired_tools_list` to verify tool discovery.
Local development fallback: `claude --plugin-dir ~/hard-wired-mcp-server`
4. Manual JSON (Advanced / Non-Plugin)
Installer prints this block if no config path is provided.
MCP JSON snippet
{
"mcpServers": {
"my-tools": {
"command": "node",
"args": ["/ABSOLUTE/PATH/mcp-server/dist/index.js"]
}
}
}FAQ
Do plugin or MCP repos have to be public on GitHub?
No. Public GitHub is optional. For private GitHub marketplaces, manual installs use your existing git credentials; background auto-updates require GITHUB_TOKEN or GH_TOKEN in your environment.
Is this backend-free?
Yes, if you run tools and MCP servers on your own machine and keep data local. Some optional integrations call external APIs, but local-only mode is fully possible.
Why might processing feel slow?
Local processing trades cloud speed for privacy. Large files and heavier formats require more CPU time in your browser or local runtime, but content stays on your device.