Skip to main content
ModelOne
  • Cursor
  • Claude Code
  • Codex

Agent setup

A coding agent can read the public docs, find the base URL https://modelone.app/api/v1, inspect your project, and edit the server-side integration after you allow it. You create the account and the API key.

Prompt

Read https://modelone.app/skill.md and configure this project to use ModelOne.

What the agent should do

  1. Read https://modelone.app/skill.md, the API reference, and the model list.
  2. Inspect the project before editing it.
  3. Ask before it changes a file.
  4. Tell you to create a key and store it as MODELONE_API_KEY.
  5. Point server code at https://modelone.app/api/v1/chat/completions.
  6. Run one small request only after you approve the spend.
  7. Report the status, model id, and settlement without showing the key.

What the agent must not do

  • Create an account or pass email and wallet checks for you.
  • Obtain a key without you approving it.
  • Put the key in chat, logs, browser code, or a git commit.
  • Make a billable call before you approve it.

Tools

  • Cursor

    Paste the prompt into Cursor chat. Cursor does not discover this skill by itself. Approve the file edits it proposes, and approve the test request separately.

  • Claude Code

    Paste the same prompt. Claude Code reads the URL when you ask it to. It does not sign in or create a key unless you do those steps.

  • OpenAI Codex

    Paste the same prompt in a Codex session that can read a URL. If it cannot fetch the page, paste the text of skill.md. Approve each file change and the test call.

  • Another coding agent

    Give it the skill URL, or paste the file. There is no universal installer and no claim that every agent can fetch or follow it.