TalinoTalinoDocumentationSource
Talino is a local macOS app. The bundled pkwn daemon stays on your machine; provider authentication remains subject to each vendor's account terms.

Talino documentation

A field guide for setting up Talino's native workspace, authorizing the coding backend you already use, and enabling project tools only when they are useful.

macOS arm64Native Electron workspace
3 backendsClaude, Codex, Gemini
Local by defaultProjects, audio, sessions

Install Talino

Talino ships as an Apple Silicon DMG. Download the current Talino-0.1.0-arm64.dmg release, open it, and drag Talino.app to Applications.

  1. 1

    Get the DMG

    A source checkout can build the same package with npm run dist; Electron Builder writes the result to release/.

    Open project releases
  2. 2

    Approve the first launch

    The current build is unsigned. If macOS blocks it, Control-click the app, choose Open, then confirm once.

  3. 3

    Choose a project folder

    Launch Talino, open a repository folder, then create or resume its session from the project rail.

Connect a backend

Talino starts its bundled pkwn daemon locally. Authorize Claude, Codex, or Gemini in Settings, then choose the connected provider and model when you create a session.

Open Settings from the account menu. Confirm the daemon reads Connected, select a provider, and start sign-in.

Finish authorization in the browser. Claude may ask for a browser code; Codex and Gemini complete through a local callback.

Select the provider and model in the session composer. Talino loads available models after the account connects.

CLI setup

For a terminal-first setup or direct daemon configuration, initialize pkwn before opening Talino.

npm install -g punakawan && pkwn init

Voice assistant

Speak to the active project

The microphone control opens the assistant. It uses on-device Whisper recognition and local speech synthesis; the selected model downloads and warms on first use.

  1. 1

    Allow Microphone access

    If it was denied, enable Talino in System Settings → Privacy & Security → Microphone, then reopen the assistant.

  2. 2

    Select recognition and voice

    Choose a Whisper tier and Kokoro voice. Apple Silicon uses WebGPU when available; use a lighter tier on CPU for quicker transcription.

  3. 3

    Choose Chat or Build deliberately

    Chat is read-only. Build creates an edit-permitted session for the active project; review its changes afterward.

Recognition and speech synthesis run locally. The coding backend receives only the request Talino sends after transcription.

Note taker

Capture the meeting without inviting a bot

The notebook control records your microphone and, with permission, meeting audio playing through your Mac. Whisper transcription and optional speaker identification run locally.

  1. 1

    Grant Microphone access

    Enable Identify speakers before recording if you need local participant clustering; its recognition model downloads on first use.

  2. 2

    Enable Screen Recording for meeting audio

    Grant Screen Recording in System Settings → Privacy & Security → Screen Recording when you need speaker audio capture.

  3. 3

    Record, label, then summarize

    Export the transcript as Markdown or summarize it with your connected coding backend.

Raw capture and transcription stay on-device. Only an optional generated summary is sent to the backend you select.

Media library

Talino keeps media jobs, assets, and prompt lineage under .talino/media in the active project.

Wan video

Talino delegates video generation to the official Wan CLI. Credentials remain in that CLI's own configuration.

  1. 1

    Install the Wan CLI

    npm install -g @wan-ai/cli
  2. 2

    Configure the Media provider

    In Settings → Media provider, select the Wan region, authenticate, set the approval policy and budget, then test the connection.

Gemini images

Add a Gemini API key and select an image model in Settings → Media provider. Talino confirms the key without storing it in the project.

Project integrations

GitHub or GitLab

Connect the host in Settings → Accounts. Use Git to inspect diffs, stage changes, create commits, and create or merge pull requests.

Deployment

In Deploy, select Vercel or Netlify, then supply a personal token or use the provider login. Talino can check, install, and run the corresponding CLI.

Session history

Search pkwn's saved transcript index from the top bar. Persistent sessions survive restarts and can be resumed from the project rail.

pkwn daemon

Talino connects to pkwn on 127.0.0.1:8787. The daemon keeps credentials and durable session transcripts in its own SQLite stores under ~/.pkwn/; Talino keeps only its interface preferences in local storage and project media under .talino/media.

Need terminal, API, gateway, automation, or VPS setup? Those are pkwn capabilities. Read the daemon documentation rather than changing Talino's local loopback setup.

Read pkwn documentation

Security and privacy

Keep pkwn loopback-bound unless you intentionally operate it behind authenticated infrastructure. Review generated changes before release, use the least-permissive agent setting that completes the task, and treat provider authorization as an account-level decision.

Talino does not add a separate metered API key. Provider behavior and consumer-account terms remain vendor-controlled.

Ready to work?

Open Talino, select a local repository, choose a connected backend, and make the first session.

Return to product overview