Set up with your coding agent

Hand the Configure integration to the coding agent you already use (Claude Code, Codex, Cursor, Devin). The setup skill at https://configure.dev/skill.md walks an agent through the whole Quick Start; you approve one browser sign-in and review the diff.

All docs pages, indexed for agents: https://docs.configure.dev/llms.txt (add .md to any page URL for plain markdown).

The one paste

Give your agent this line in your app's repo:

Read https://configure.dev/skill.md and follow it to add Configure to this app.

That URL is the machine-readable setup skill. Any agent that can fetch a URL and run shell commands can follow it end to end. The paste is the whole instruction: the skill contains every command in its exact order, so do not hand your agent a step list from this page; an agent should execute the skill, not reconstruct it.

What you will see happen

Reading the skill, the agent drives everything through the configure CLI, never a dashboard. Expect this sequence:

1. npm install configure and npx configure setup --users, which creates the API keys and OAuth client and writes the five credentials to .env.
2. npx configure add callback, which generates the callback route, server-side code exchange, and sign-in button for the local callback registered by setup.
3. Configure OAuth appears beside your existing sign-in providers, and a place for users to manage permissions after sign-in.
4. npx configure verify proves the integration, then a real configure_profile_read tool call through your tool router proves it end to end. (configure_profile_read is a tool the model calls inside a chat turn, not a CLI command; there is no npx verb for it.)

Done means a chat turn that runs the loop: the model calls configure_profile_read at the start and the reply uses the returned profile, read securely through the Configure server; Configure tools answer during the turn; and a commit after the turn writes back what it learned. A rendered button does not count; the skill holds the agent to the tool calls.

Your one step

When setup opens the browser, approve the Configure developer sign-in and choose or create the agent handle for your app. Everything else is the agent's job; the skill forbids asking you to copy keys from a dashboard.

No account, and you want a dry run first? The skill can provision Test mode (sandbox) instead: test keys plus a synthetic user with a full profile, no signup.

After setup, the agent drives the account

The CLI covers the developer dashboard, so keep asking the agent instead of clicking:

npx configure login                 # one-time browser approval
npx configure whoami                # which account it is acting as
npx configure users                 # who signed in to the app
npx configure conversations         # what the agent committed lately
npx configure keys                  # list or create keys
npx configure add origin URL        # register a deployed callback
npx configure logs tail --failed    # follow failures live while testing
npx configure usage                 # reads and users against the plan
npx configure billing               # plan and renewal

The skill knows these docs

The skill points the agent at https://docs.configure.dev/llms.txt, the agent-readable index of every page here. Before wiring anything, the agent reads the pattern that fits your product: Choose your integration for the four shapes, Handling Tool Calls for the model loop, Configure MCP when your runtime speaks MCP.