Frequently asked questions

Yes. Hope Agent is open-source under the MIT license, so you can freely use, modify, and distribute it. The only thing you might pay for is API usage from your chosen model provider (unless you run a local model).

macOS is the primary desktop platform. The v0.60.0 Release includes both Apple Silicon arm64 and Intel x64 DMGs; choose the installer matching your chip. Windows x64 and Linux x64 / arm64 support is currently experimental. Check Releases for the actual assets of the version you need.

Configuration, chats, memory, attachments, skills, and logs stay under ~/.hope-agent/ by default. Model requests go directly to your configured provider. Cloud models receive the relevant inputs, and external tools use the data needed for their calls. Server mode uses an Owner Token and protected browser sessions, with SSRF checks for web fetching.

Use a provider API key, sign in with Codex OAuth, or run a local model without a cloud API key. First install the runtime manually from Ollama, then return to Hope Agent Settings to download and select a model. Automatic Ollama installation is currently disabled.

Run these two commands in Terminal to strip the quarantine attribute and re-sign the app ad-hoc, then reopen it: sudo xattr -cr /Applications/Hope\ Agent.app and sudo codesign --force --deep --sign - /Applications/Hope\ Agent.app.

This means the system is missing the required runtime. Install the Microsoft Visual C++ 2015–2022 Redistributable (x64) and relaunch the app (Windows is x64-only for now).

The desktop updater prefers the repo.hopeagent.ai mirror and falls back to GitHub Releases. Check through Settings → About or chat; downloaded updates are signature-verified. Package-manager installs can use the same updater. Docker deployments upgrade by pulling a new image and recreating the container with the existing data volume.

Yes. Run hope-agent server start to launch an HTTP/WS daemon, and hope-agent server install registers it as a launchd / systemd auto-start unit so it lives long-term on a NAS, cloud VM, or spare laptop. The server ships an embedded Web GUI — open http://<server>:port from any phone, tablet, or browser to get the full UI; an official Docker image (ghcr.io/shiwenwen/hope-agent) is also available for one-command self-hosting.

Desktop GUI is the most complete default entry point and also embeds an HTTP/WS server; Server mode (server start) is a headless daemon with an embedded Web GUI, ideal for 24/7 uptime running IM channels and cron jobs; ACP mode (acp, over stdio) plugs Hope Agent into ACP-capable IDEs as an agent backend. All three share the same ha-core core and store data under ~/.hope-agent/.

If you already installed Hope Agent.app manually, append --adopt to brew install --cask hope-agent to let the cask take over your existing same-version app without re-downloading, or --force to force a fresh download and overwrite it.

The container generates an Owner Token. Read it with docker exec hope-agent hope-agent server token show (or docker compose exec hope-agent hope-agent server token show), then enter it in the browser login to obtain an HttpOnly session. Keep the Root Token out of URLs and localStorage. Use HTTPS for public access; a VPN does not replace the token.