Supported environment
- Operating system
- macOS
- Browser
- Google Chrome
- Local tools
- Python 3.11 or newer and Git
- Agent
- An AI agent or app that supports a local MCP connection
Brave, Edge, Arc, Firefox, Windows, Linux, mobile, and Safari are not current support promises.
Install the two local components
1. Companion local service
Clone the public repository and run the installer from the project directory:
git clone https://github.com/AadityasinhJadeja/Daemon-Mode.git
cd Daemon-Mode
./tools/install-local-service-launch-agent.shThis installs and starts the companion service for your macOS user. It does not delete or replace an existing memory vault.
The installer registers a per-user macOS LaunchAgent. The service starts after you log in, returns after a normal reboot once you log in again, and is restarted by macOS if the process exits unexpectedly. The dashboard is only a view into the service; closing its tab does not stop capture.
2. Chrome extension
Open chrome://extensions, enable Developer mode, choose Load unpacked, and select the extension directory from the downloaded project.
Installing only the extension does not create or host the memory vault. Chrome captures allowed pages; the companion service stores and retrieves them.
If the service is off
The background service does not run before login or while your macOS user is logged out. After login, the extension should reconnect without a dashboard or terminal window.
- Open the Daemon extension and confirm it reports that the companion service is unavailable.
- From the project directory, run the installer again. The installer verifies the managed files, reloads an unloaded service, and preserves the memory vault.
- Confirm the loopback health endpoint responds:
curl --fail --silent --show-error http://127.0.0.1:4317/healthIf the service still does not start, share only the error text in a support request.
Connect an MCP client
The setup helper prints machine-correct configuration without changing anything by default:
python3 tools/setup-agent-harness.pyWhen ready, explicitly connect one verified client:
python3 tools/setup-agent-harness.py --apply codex --scope userReplace codex with claude-code, cursor, or claude-desktop. The helper backs up existing configuration, preserves unrelated settings, and refuses malformed or conflicting files.
ChatGPT is not configured by this local helper. Its integration path is remote MCP/apps/connectors and would require a future bridge that preserves Daemon's non-destructive and protection boundaries.
Update safely
After downloading the latest changes, run the local service installer again. It verifies the new release, preserves the database, activates the update atomically, and restores the prior runtime if startup fails.
./tools/install-local-service-launch-agent.shThen return to chrome://extensions and reload Daemon.
Uninstall without accidental data loss
Stop the background service with:
./tools/uninstall-local-service-launch-agent.shThis preserves the managed runtime and SQLite memory by design. Removing the Chrome extension stops future capture but also does not delete the separate SQLite vault. Use the human-owned dashboard controls to export or clear memory before deleting files manually.
Ask for help safely
Use example data and the smallest error message needed to explain the problem. Keep browsing content, credentials, databases, and local configuration out of public issues.
Use GitHub Issues for installation or product problems. For a possible vulnerability, follow the Security reporting guidance.