Set Up PaPut for the First Time
Connect PaPut to your AI client and run onboarding to prepare memo candidates and the foundation of your skill sheet from past work. If you have already completed part of the setup, onboarding resumes from what is still missing.
1. Install and connect PaPut
In Claude Code, claude.ai, Claude Desktop, and Codex, install the PaPut plugin. In clients without plugin support, such as ChatGPT, configure https://mcp.paput.io/mcp as an MCP connector.
The OAuth authorization screen appears the first time you use a PaPut tool. Log in to PaPut and allow the connection.
2. Ask your AI to start onboarding
Ask your AI to "guide me through the initial PaPut setup." In Claude Code, you can also invoke the skill directly.
/paput:onboardingEven in a client without skill support, your AI can follow the initial-setup guidance included in PaPut tool responses when the client exposes it.
3. Complete two confirmations
The first confirmation allows your AI to read past sessions stored locally. If you agree, it reviews up to 10 of the newest sessions and extracts memo candidates together with material for your skill-sheet foundation. In a client that cannot read local files, it skips this step and continues with the information available to it.
The second confirmation is a single review before saving. Your AI presents a manifest of memo candidates and drafts for your basic information, skills, and projects. Check whether each memo should be public or private and which project it belongs to, then approve the items to save. You can exclude individual items by number.
4. Keep capturing knowledge
After saving your initial data, your AI proposes adding the PaPut usage rules. If you agree, it runs the following command so reusable knowledge candidates can be captured during future work.
npx -y paput-mcp setup-ai --rules-onlyWhen setup finishes, your AI provides your public profile URL. You can then continue with project episodes, a self-PR, dashboard analysis, or harvesting the rest of your past sessions.