BCI-MCP

Live brain state · plug your brain into any AI

First 15 minutes: try BCI-MCP without a headset

Use a synthetic signal to explore the dashboard, neurofeedback, recording, and replay.

  1. Start the synthetic dashboard. In a terminal, run:
    bci-mcp dashboard --device synthetic://

    Open http://127.0.0.1:8000. If you started this dashboard with another device, restart it with the command above to follow along.

  2. Watch the bars. After a short warm-up, the metric bars and relative frequency-band columns below update as the synthetic signal changes. The columns are scaled relative to the largest band, not an absolute EEG measurement.
  3. Try neurofeedback. Leave the dashboard running and open a second terminal. Run a 30-second focus session:
    bci-mcp neurofeedback --device synthetic:// --metric focus --target 0.7 --seconds 30

    Watch the current score and time in the target zone, then read the summary. This uses a separate synthetic stream; it does not control the dashboard.

  4. Record a sample. In the second terminal, run:
    bci-mcp record --device synthetic:// --seconds 30 --out session.npz

    This records a new synthetic stream to session.npz in that terminal's working directory, not the dashboard stream.

  5. Replay it. From the same directory, run:
    bci-mcp play session.npz

    The recording loops in the terminal brain-meter. Press Ctrl+C to stop; use Ctrl+C in the dashboard terminal to stop it too.

Synthetic data is simulated, not your brain activity. Focus and calm are heuristic band-power proxies, not SSVEP classifiers or clinical measurements.

Concentration ratio

Concentration ratio = beta power ÷ (alpha power + theta power). The current ratio is waiting for signal….

The focus score below maps this ratio to a value between 0 and 1; it is not the raw ratio. More beta relative to alpha and theta tends to raise the score. Moving the synthetic slider changes the signal used in this calculation.

This is a heuristic proxy, not a direct measurement of concentration or a clinical test. Synthetic data does not represent your brain activity.

Heuristic band-power proxies — not clinical, diagnostic, or event-level EEG. Welch PSD averages transients out (no ERPs, spindles, or bursts). Weight readings by confidence and treat an unreliable signal as untrustworthy.
Brain wave glossary

These names group waves by how quickly they repeat; Hz means repeats per second. The ranges below are the ones used by this dashboard; other sources may use slightly different boundaries.

No band on its own tells you what someone is thinking or feeling. The synthetic signal is generated by the computer, not recorded from a person.