HiveMind Flask Chatroom

June 25, 2026 · View on GitHub

logo

A multi-user Flask chatroom that connects to a HiveMind hub as a single satellite and fans it out to many browser users. Each visitor chats under their own username and language; messages are routed to your OpenVoiceOS assistant through the hub and spoken replies are streamed back into the room.

chatroom

Where it sits

HiveMind is a mesh: satellites connect to a central hivemind-core hub over an authenticated, encrypted protocol. This app is one satellite — it holds a single set of HiveMind credentials and authenticates server-side using the Python hivemind-bus-client. Browser users never see or supply credentials; they just pick a name and talk.

many browsers ──HTTP──► Flask app (1 HiveMind credential) ──encrypted──► hivemind-core ──► OVOS / agent

This is the counterpart to HiveMind-webchat, which connects client-side from each browser instead. Use this when you want a shared room where the credential lives on the server.

Install

pip install hivemind-flask-chatroom

Or from a checkout:

pip install .

Runtime dependencies: flask, hivemind-bus-client (the bus-client 2.x line), ovos-bus-client, ovos-utils. Packaging is driven entirely by pyproject.toml — there is no setup.py or requirements.txt; the version is read from hivemind_chatroom/version.py by the shared OpenVoiceOS release workflows.

Quickstart

1. Run a hub and add this client

On the machine hosting the assistant, install and run hivemind-core:

hivemind-core add-client      # prints an access key + password
hivemind-core listen --port 5678

2. Provision the chatroom's identity

The chatroom authenticates from the HiveMind identity file rather than CLI flags. Write the credentials from the previous step into it with the hivemind-bus-client CLI:

hivemind-client set-identity \
  --key <access-key> \
  --password <password> \
  --host 127.0.0.1 --port 5678 \
  --siteid flask

This populates ~/.config/hivemind/_identity.json, which the app reads on startup. Verify it can reach the hub with hivemind-client test-identity.

3. Start the chatroom

hivemind-flask-chatroom --port 8985

Open http://localhost:8985. You are redirected to a room as anon_user; to join under a specific name, language, or site, visit /chatroom/<username>/<site_id>/<lang> directly (for example /chatroom/alice/livingroom/en). Type a message and the assistant's reply appears prefixed with your username.

Command-line options

usage: hivemind-flask-chatroom [-h] [--port PORT] [--host HOST]
                               [--log-level LOG_LEVEL] [--log-path LOG_PATH]

options:
  -h, --help            show this help message and exit
  --port PORT           Chatroom port (default 8985)
  --host HOST           Bind address (default 0.0.0.0)
  --log-level LOG_LEVEL DEBUG / INFO / ERROR (default DEBUG)
  --log-path LOG_PATH   Log directory; default <xdg_state>/hivemind,
                        or "stdout" to log to the console

How it works

  • On startup MessageHandler.connect() opens one HiveMessageBusClient (useragent JarbasFlaskChatRoomV0.2, site_id="flask") to the hub and subscribes to speak, ovos.common_play.play, and the legacy audio-play message.
  • Each browser user maps to an OVOS Session keyed by username, so per-user lang and site_id preferences persist across turns. Outgoing utterances carry that session so skills can tell users apart.
  • Routes: GET / redirects into a room; GET /chatroom/<username>/<site_id>/<lang> renders the room; POST /send_message submits an utterance; GET /messages returns the running message log as JSON (the page polls it).

The HiveMind link is end-to-end encrypted between the Flask process and the hub; the browser-to-Flask hop is plain HTTP, so front it with a TLS reverse proxy (nginx, Caddy) for any non-local deployment.

Running the tests

pip install -e ".[e2e]"   # unit + e2e deps (resolves the bus-client 2.x stack)
pytest tests/             # everything
pytest tests/test_smoke.py   # unit smoke tests only (no network)
pytest tests/e2e/            # real-hub end-to-end suite

The end-to-end suite boots a real hivemind-core master in-process via the hivescope harness and connects the real chatroom over a real HiveMessageBusClient; only the Flask browser surface is mocked. See docs/development.md.

See also

  • docs/usage.md — identity provisioning, multi-user routing, and the message API in more detail.
  • docs/development.md — packaging, the test layout, and how the e2e harness wires a real hub to the real chatroom.
  • HiveMind-webchat — the browser-side single-user equivalent.

License

Apache 2.0 — see LICENSE.