How to Disable the llmfit Web Dashboard

Pass the --no-dashboard flag to the llmfit CLI binary to suppress the automatic background web server startup.

The llmfit tool from the AlexsJones/llmfit repository provides an interactive TUI with an optional web dashboard. By default, the binary launches a background HTTP server to serve the UI, but you can disable this behavior using a simple command-line argument to improve security or reduce resource usage.

Understanding the Automatic Dashboard Behavior

When you run llmfit without flags, the application automatically attempts to start a web dashboard. According to the source code in llmfit-tui/src/main.rs, the binary checks environment variables for the dashboard host and port, verifies if the server is already reachable, and starts the HTTP server if needed. It also writes a PID file for later cleanup. This entire flow is gated by the CLI configuration and executes unless explicitly disabled.

Disabling the Dashboard with --no-dashboard

The primary method to disable the llmfit web dashboard is the --no-dashboard flag. This boolean argument is parsed into the no_dashboard field of the internal Cli struct. The startup logic at lines 180 and 2779 of llmfit-tui/src/main.rs evaluates !cli.no_dashboard, skipping the server initialization when the flag is present.

CLI Usage Examples

Run the binary without the dashboard:


# Default behavior - dashboard starts automatically on http://<host>:<port>/

llmfit

# Disable the web dashboard completely

llmfit --no-dashboard

Python Wrapper Usage

If you invoke llmfit through the Python module wrapper, forward the flag directly after the module name:

python -m llmfit --no-dashboard

Technical Implementation Details

The dashboard disable logic is implemented in the Rust source code with precise conditional checks. In llmfit-tui/src/main.rs, the decision to start the server depends on the evaluation of !cli.no_dashboard. When this condition returns false (meaning the user provided the flag), the code bypasses the server initialization routine defined in llmfit-tui/src/serve_api.rs.

The user-facing documentation in docs/tui.md (line 322) confirms this flag prevents the automatic launch of the dashboard server, providing authoritative confirmation of the behavior described in the source.

Summary

  • The --no-dashboard flag completely disables the automatic web dashboard startup in llmfit
  • When omitted, the binary binds to the configured host/port and manages a PID file for the background process
  • The flag functions identically whether using the native Rust binary or the Python wrapper
  • Implementation resides in llmfit-tui/src/main.rs with the HTTP server logic contained in llmfit-tui/src/serve_api.rs

Frequently Asked Questions

What port does the llmfit dashboard use by default?

The dashboard server binds to the host and port specified via environment variables. When reachable, it serves the interactive UI on http://<host>:<port>/. Check your environment configuration or the startup logs for the specific binding address, as the defaults are determined by the application's configuration parser.

Can I disable the dashboard if I'm using the Python wrapper?

Yes. Pass --no-dashboard after the module invocation: python -m llmfit --no-dashboard. The Python wrapper forwards arguments directly to the underlying Rust binary's CLI parser, ensuring the no_dashboard field is set correctly in the Cli struct.

Does disabling the dashboard affect the TUI functionality?

No. The terminal user interface operates independently of the web dashboard. Disabling the web server only prevents the background HTTP process from starting and writing its PID file; the interactive TUI continues to function normally within your terminal session.

Where is the dashboard server implementation located?

The HTTP server implementation resides in llmfit-tui/src/serve_api.rs. The startup decision logic that checks the no_dashboard field before invoking the server is implemented in llmfit-tui/src/main.rs at lines 180 and 2779.

Have a question about this repo?

These articles cover the highlights, but your codebase questions are specific. Give your agent direct access to the source. Share this with your agent to get started:

Share the following with your agent to get started:
curl -s "https://instagit.com/install.md"

Works with
Claude Codex Cursor VS Code OpenClaw Any MCP Client

Maintain an open-source project? Get it listed too →