# WitSelf > Memory that moves with your AI. Give your AI agents identity, facts, and memory across sessions, tools, and machines. Install WitSelf, then manage it through your AI in plain English. ## Keep the context. Change the tool. Decisions, preferences, and project history belong with your agent. WitSelf makes that context available through the CLI and the Model Context Protocol (MCP), wherever the same agent is connected. ### Remember what matters Save precise facts and the context behind them. Recall decisions and lessons in the next session. ### Bring your own AI Use your existing AI tools and model subscriptions. Your AI does the reasoning; WitSelf stores and retrieves its context. ### Grow into a team Paid plans add encrypted secrets, durable agent messaging, and collaboration. Team adds more isolated workspaces, called realms. ## Look inside your agent. Explore facts, memories, and transcripts. See recorded activity at a glance. Open the local Agent Console in your browser, or stay in your terminal. ### Agent web console A browser view of your agent’s context and activity, served locally by WitSelf on your computer. ![WitSelf local Agent Console showing a saved decision, its supporting evidence, and version history with synthetic demo data.](https://witself.com/assets/screenshots/agent-console.png) Open your own agent: ```sh witself dashboard serve --account personal --realm main --agent assistant --open ``` Starts the local console and opens your browser. Keep the command running while you use it. [Read the guide](https://github.com/witwave-ai/witself/blob/main/docs/agent-console.md). ### Terminal workspace Your agent’s data in your terminal, with keyboard navigation and live refresh. ![WitSelf terminal workspace showing a list of memories alongside a selected decision and its supporting evidence, with synthetic demo data.](https://witself.com/assets/screenshots/agent-tui.png) Open your own agent: ```sh witself tui --account personal --realm main --agent assistant ``` Loads your agent’s data. Press b to open its local web console in your browser. [Read the guide](https://github.com/witwave-ai/witself/blob/main/docs/agent-tui.md). Real product interfaces, shown with synthetic demo data. Select either image for a closer look. After setup, use your account, realm, and agent names. These examples match personal / main / assistant in the setup guide. Both views connect to your agent’s stored data; manage WitSelf through your AI or the CLI. Just exploring? Try demo data without an account: ```sh witself tui --demo ``` ## Supported integrations Commands assume a configured agent. - **Codex**: MCP + transcript hooks. macOS, Linux, Windows. Command: `witself install codex`. - **Claude Code**: MCP + transcript hooks¹. macOS, Linux, Windows. Command: `witself install claude`. - **Grok Build**: MCP + transcript hooks¹. macOS, Linux, Windows. Command: `witself install grok`. - **Cursor**: MCP + transcript hooks. macOS, Linux; Windows via WSL. Command: `witself install cursor`. - **DeepSeek Harness**: MCP + transcript hooks¹. macOS, Linux, Windows. Command: `witself install dsh`. - **OpenClaw**: Preview · MCP + routing. macOS, Linux, Windows. Command: `witself install openclaw`. - **Antigravity**: Preview · MCP + routing. macOS, Linux, Windows. Command: `witself install antigravity`. - **GitHub Copilot CLI**: Preview · MCP + routing. macOS, Linux, Windows. Command: `witself install copilot`. ¹ On native Windows, Claude Code, Grok Build, and DeepSeek Harness install MCP and routing without transcript hooks. Cursor requires WSL on Windows. Preview integrations do not include WitSelf transcript hooks or automatic prompt-context injection. OpenClaw preview requires one configured default agent. Support varies by client and version; an installer is not a guarantee of every model workflow. ## Managed-service pricing Flat monthly pricing in USD. AI provider subscriptions and inference costs are separate. | Plan | Personal | Professional | Team | | --- | --- | --- | --- | | Monthly USD | $0 | $30 | $250 | | Realms (isolated workspaces) | 1 | 1 | 25 | | Agents per realm | 10 | 100 | 100 | | Active memories per agent | 1,000 | 10,000 | 50,000 | | Facts per agent | 1,000 | 10,000 | 50,000 | | Secrets per agent | Not included | 100 | 250 | | Transcript retention | 30 days | 90 days | 365 days | | People on the account | 1 | 3 | 25 | | Agent messaging & collaboration | Not included | Included | Included | | Support | Not included | Included | Included | Enterprise is coming soon. Agent email, email aliases, and custom email domains are coming soon and are not currently delivered features. Full catalog: [plans.json](https://witself.com/plans.json). Catalog updated: 2026-10-10. ## Manage WitSelf through your AI Your words. Your AI. Your WitSelf. Install once. Manage it by asking your AI. Ask the AI you already use to check your plan, explain an update, or help solve a problem. It works through the installed WitSelf CLI. No separate web management dashboard to learn. ### Stay up to date “Check my WitSelf version, explain what changed, and help me update.” ### Understand your plan “What’s included in my plan, and what would Professional add?” ### Get help “Help me troubleshoot this. If we need support, prepare a ticket for me.” ### Example conversation: From free to Professional **You:** I’m ready for Professional. What changes? **Your AI:** Professional adds more memory, encrypted secrets, and agent collaboration. It’s $30 USD per month, billed monthly. Shall I start the upgrade? **You:** Yes, go ahead. **Your AI:** Here’s your secure checkout. Once you finish, I’ll check that Professional is active. Illustrative first paid upgrade. Your AI checks the current plan and price before making a change. You approve the purchase and complete checkout when required. Use an AI assistant with permission to run local commands and access your WitSelf account. A chat connected only through MCP can use agent tools, but cannot perform these account-management steps. Ask your AI questions on any plan. Professional and Team also include support tickets; Personal includes public documentation and community resources. ## Install and connect Ask this to have AI help you get everything set up: “Read https://witself.com/product.md, then help me install WitSelf, set up my free account, and connect it to this AI. Walk me through any steps I need to complete myself, then verify everything works.” Use an AI with local command access. You’ll complete identity verification and review the terms. ### Homebrew · macOS / Linux ``` brew install witwave-ai/tap/witself ``` ### Shell · macOS / Linux ``` curl -fsSL https://raw.githubusercontent.com/witwave-ai/witself/main/install.sh | sh ``` ### PowerShell · Windows x64 ``` [Net.ServicePointManager]::SecurityProtocol = [Net.SecurityProtocolType]::Tls12 irm https://raw.githubusercontent.com/witwave-ai/witself/main/install.ps1 | iex ``` ### Create your account and agent Start free. Complete the browser challenge, retry with the challenge token, and verify your email. Then create a realm, agent, and local token. ```sh witself account create --name personal --email you@example.com --accept-terms # Complete the challenge; retry the same command with --challenge TOKEN. # Follow the verification link in your email, then check status: witself account status --account personal witself realm create --account personal main # Use the returned realm ID and agent ID in the next commands: witself agent create --account personal --realm REALM_ID assistant witself token create --account personal --agent AGENT_ID ``` ### Connect your AI tool Install your preferred integration, restart the AI client, and start a new session. Then ask it to remember something useful. The example below connects Codex with user-scoped hooks. ```sh witself install codex --account personal --realm main --agent assistant --user-hooks witself integrations --verify ``` Replace you@example.com, REALM_ID, and AGENT_ID with your own values. Read the [Terms](https://self.witwave.ai/legal/terms) and [Privacy Policy](https://self.witwave.ai/legal/privacy) before using --accept-terms. Release installers verify checksums. ## Questions ### Is WitSelf an AI model? No. WitSelf is the persistent context and identity layer for your agents. Your connected AI client performs the reasoning. Model subscriptions and inference costs stay with that provider. ### Do I need a web management dashboard? No. Manage WitSelf through the CLI, directly or by asking your AI to use it. There is no account sign-in on this product page. Signup verification and secure payment setup may open a browser. ### Can my AI update WitSelf and get support? An AI with local command access can check your installed version, explain release notes, and help run the appropriate installer or package-manager update. It can also help troubleshoot and, on Professional or Team, open and follow support tickets through the CLI. Your AI guides the work; support tickets go to the WitSelf team. ### Can I move between AI tools? Yes. Connect supported tools to the same WitSelf account, realm, and agent to use that agent’s saved facts and memories. Transcript capture and automatic context differ by integration. ### Can I run it myself? The source and self-hosting documentation are on GitHub. The plans above describe the managed service; self-hosted infrastructure and operations are your responsibility. ## Links - [Source repository](https://github.com/witwave-ai/witself) - [Documentation](https://github.com/witwave-ai/witself/tree/main/docs) - [Feature status](https://github.com/witwave-ai/witself/blob/main/docs/feature-status.md) - [CLI reference](https://github.com/witwave-ai/witself/blob/main/docs/cli-command-surface.md) - [Witwave](https://witwave.ai/) WitSelf is a product of Witwave.