Skip to content
mordn

Build with agents

Hand mordn to your agent. Skip the backend work.

Point your coding agent at one skill file. It detects your framework and existing auth, scaffolds the route, wires identity from your verified session, and mounts a production-ready chat agent — secure by default, not bolted on after.

Paste into your agent
  • Claude Code
  • Cursor
  • OpenAI Codex
  • GitHub Copilot
  • Windsurf
  • Gemini CLI
Add @mordn/chat-widget to this app.

Read https://www.mordn.com/SKILL.md and follow it end to end:
- Detect my framework and existing auth provider.
- Scaffold the route at app/api/chat/[[...chat]]/route.ts.
- Wire getUserId from my verified server session (return null when unauthenticated; never read identity from a header or body).
- Mount <ChatWidget apiBase="/api/chat" /> and import the prebuilt styles.
- Set the required env vars (MORDN_CHAT_KEY for the hosted handler).
- Use the hosted handler to start. Don't invent credentials or widen my public routes.
- Run the build and verify /api/chat/bootstrap returns 200 for an authenticated request.
Reads mordn.com/SKILL.md

Works with

The coding agents you already use.

Any agent that can fetch a URL and edit files can run the install. If yours isn't listed, the skill file is plain Markdown — point any agent at it.

  • Claude Code

    Anthropic's terminal-native coding agent

  • Cursor

    The AI-first code editor

  • OpenAI Codex

    OpenAI's cloud software engineer

  • GitHub Copilot

    GitHub's in-editor assistant

  • Windsurf

    Codeium's agentic IDE

  • Gemini CLI

    Google's CLI coding agent

Why this is different

An install agent can't break the security model.

mordn's identity boundary is server-side and non-negotiable. The skill file encodes that boundary as hard rules — your agent wires it correctly or it stops and reports.

  • Identity stays server-side

    The prompt tells your agent to wire getUserId from your verified session. The browser never tells the server who it is — the security model is preserved, not bypassed.

  • Reads one skill file

    SKILL.md is a self-contained install skill: detect framework and auth, scaffold the route, mount the widget, set env, verify the build. No doc crawling required.

  • Fails closed

    The prompt forbids inventing credentials or widening your public routes. If the agent can't verify identity, it stops and reports — it doesn't paper over the boundary.

How it works

One prompt. One skill file. A working agent.

Your agent fetches SKILL.md, which is self-contained: detection, scaffold, wire, mount, verify. No manual steps in between.

  1. 01

    Paste the prompt

    Copy the prompt from above into your coding agent. It points at mordn.com/SKILL.md — the full install skill.

  2. 02

    Agent reads + scaffolds

    SKILL.md tells the agent to detect your framework and auth, scaffold the route, wire getUserId from your verified session, and mount the widget.

  3. 03

    Build + verify

    The agent sets your env vars, runs the build, and confirms /api/chat/bootstrap returns 200 for an authenticated request. Then it reports back.

The prompt

Copy this. Hand it to your agent.

It points at SKILL.md, the self-contained install skill. Prefer the raw file? Fetch /SKILL.md directly, or grab the full agent reference.

  • mordn.com/SKILL.md
  • mordn.com/llms-full.txt
  • mordn.com/llms.txt
Paste into your agent
  • Claude Code
  • Cursor
  • OpenAI Codex
  • GitHub Copilot
  • Windsurf
  • Gemini CLI
Add @mordn/chat-widget to this app.

Read https://www.mordn.com/SKILL.md and follow it end to end:
- Detect my framework and existing auth provider.
- Scaffold the route at app/api/chat/[[...chat]]/route.ts.
- Wire getUserId from my verified server session (return null when unauthenticated; never read identity from a header or body).
- Mount <ChatWidget apiBase="/api/chat" /> and import the prebuilt styles.
- Set the required env vars (MORDN_CHAT_KEY for the hosted handler).
- Use the hosted handler to start. Don't invent credentials or widen my public routes.
- Run the build and verify /api/chat/bootstrap returns 200 for an authenticated request.
Reads mordn.com/SKILL.md

Add an agent to your app in minutes.

Open source.

Start buildingRead the docs