Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.tryscott.ai/llms.txt

Use this file to discover all available pages before exploring further.

Get a quick overview

How Scott AI works

Planspaces, context flows, and an agent-centric project lifecycle.

The Scott Method

Our philosophy on building with agents — goal setting before agent delegation.

Install

Run the following command to install the scott terminal integration. This curl command wires Scott to any detected coding agents, and opens a browser window for sign-in.
curl -fsSL https://tryscott.ai/install | sh
After install, use the Scott terminal integration in any coding agent by saying somethig like:
# In Claude Code, Codex, OpenCode, Cursor etc.
"push this conversation to Scott"
Your agent will prepare the context in a form optimized for sharing with your team, and push to Scott so that you can collaborate on the ideas with your team in the UI.Once your team is sufficiently aligned, you can pull the context back to a new session with all relevant team feedback:
# In Claude Code, Codex, OpenCode, Cursor etc.
"pull the project requirements from this planspace and start building -- https://app.tryscott.ai/workspace/xxxxxxx-8f6c-yyyyyy-9056-07d8ca0-zzzzz"
Under the hood, Scott uses skills, an mcp and a CLI to integrate with the UI. If the installer cannot detect your coding agent automatically, you can manually complete the setup with this command:
scott setup --claude   # or --cursor, --codex, --opencode, --gemini
See the full CLI reference.

Set up Scott AI

  1. Create an account — see Install above. The browser app is recommended for PMs and designers; and you can install a CLI so developers can interact from scott from their coding agent.
  2. Create a planspace — name your planspace after your next big idea that you have been thinking about
  3. Attach contextconnect your GitHub repo, upload docs, or point to a local folder so the agent can ground its output in your codebase
  4. Describe what you want to build — start with a rough prototype from your coding agent, or describe the intent from scratch
  5. Invite your team to review — when the idea is more fleshed out, request reviews from your team
  6. Iterate — as with all good ideas, iteration and collaboration refines the idea into something great
Start with a real project your team is about to begin. The best way to evaluate Scott is with a plan that actually matters.
Next steps: Sharing · Reviews · Making the most of Scott AI
  1. Read The Scott Method at tryscott.ai/method
  2. Set up for your team see I’m setting up for my team above
  3. Read measuring Scott impactPulse
  4. See how intent links to codeScott CI auto-links PRs back to approved specs
Next steps: For You · Scott CI · Making the most of Scott AI