Your Mac.
Your AI shortcut layer.
Codec is a Mac-native AI shortcut layer for power users. Hold Cmd+R, speak, paste. Press F5 for live typing. Right-click any text for 8 AI services: proofread, translate, explain, summarize. 88 skills. Vision-based UI clicking (“click the Submit button” actually works). Local LLMs (Qwen, Llama, Mistral via MLX) or cloud (GPT-4o, Claude). Whisper for STT, Kokoro for TTS. Free. MIT-licensed. Your voice. Your computer. Your rules.
$ cd codec && ./install.sh
✔ macOS 13+ · Apple Silicon
✔ Installed 88 skills · 400 features
✔ 2,000+ tests green
✔ Models running locally — nothing leaves this Mac.
# Hold Cmd+R anywhere. Speak. Release.
you> "summarise the doc on screen"
→ skill: screenshot_text
→ model: local · 0 tokens sent
→ text pasted at cursor.
$
Codec is a macOS AI shortcut layer for power users. Voice-control your Mac, dictate anywhere with Cmd+R, right-click any text for 8 AI services, run a 250K-context chat with 12 agent crews, do vision-based UI clicking. Free. MIT-licensed. No subscription. Install yourself in one command, or have AVA Digital install Codec + an LLM + voice-to-text + text-to-speech for you (personalized installation from €500).
See how it thinks.
A short tour of what Codec does: the framework behind every AVA agent, running in the open.
Five layers.
Swap any of them.
Interface
Agent
Skills
Runtime
Model
One framework. Seven products.
Codec is the runtime. Around it we ship six surfaces (voice, dictation, instant agents, chat, vibe sessions, voice-first) plus an overview surface that ties them together. All open source. All swappable. All shipped from the same monorepo.
CODEC for Mac.
A native app for the seven.
One drag-and-drop install. The full Codec suite, packaged for macOS. Bring your own model. We never resell tokens.
Native Mac app
Built for Apple Silicon. Drag-and-drop install. Menu-bar, hotkeys, system-wide voice. No browser, no Electron tax.
Pick your model
Plug in Anthropic, OpenAI, Gemini, or a local model. Switch per surface. Usage is routed at provider cost × 1.20 — we never resell tokens.
Voice-first
Push-to-talk anywhere. Dictation, conversation, voice agents. The keyboard becomes optional.
CODEC OSS
Self-host the runtime. Build your own surfaces. We don’t see your traffic.
- Codec Core, Skills SDK, evals
- Self-hosted, your cloud, your keys
- Community support · GitHub Discussions
- Fork, modify, ship commercial work on top
CODEC for Mac
+ model usage at provider cost × 1.20 (we route, never resell). Cancel anytime, OSS still yours.
- All seven surfaces, native macOS
- Auto-updates, signed builds, hotkey daemon
- System-wide dictation, voice, push-to-talk
- Bring your own model · one config switch
One Mac · one year · renews annually, cancel anytime · your licence key is emailed immediately after payment.
Why we made it open.
We spent two years watching the agent tooling space congeal into closed platforms, each of them trying to be the landlord of your workflows. We don't think that works long-term. The interesting surface area is above the runtime, not inside it.
Codec is our bet that the best agents will be built on open foundations, and the best agencies will be the ones who can run yours end-to-end. You pay us for the people and the playbook. Not the plumbing.
Your data, your box
Self-host on your cloud. We never see customer data unless you ask us to.
No vendor lock-in
Agents and skills are portable. If we disappear tomorrow, you keep running.
Fork & ship
Build your own agent on Codec. Sell it, license it, hide it. No strings.
Read our code
Every skill, every prompt, every eval. If you can't audit the agent, you shouldn't run it.
Stop renting. Own the stack.
Every Codec install is a migration out of the SaaS subscriptions that stack up every month. Same tools. Same job. No monthly bill after setup.
Install yourself.
Clone from GitHub. Run ./install.sh. A 9-step setup wizard configures your LLM, voice, hotkeys, and Google OAuth. Free. MIT-licensed. No subscription. Bring your own keys (or run fully local with MLX).
For Mac power users comfortable with a terminal.
We install it for you.
We come to your Mac (or remote-pair) and install Codec end-to-end: your choice of LLM (local Qwen / Llama / Mistral via MLX, or cloud GPT-4o / Claude), Whisper for speech-to-text, Kokoro for text-to-speech, vision UI clicking, Google Workspace OAuth, remote access via Cloudflare Tunnel or Tailscale, audit trail.
Tailored to your workflow. You own the install. We pre-configure every shortcut.
Your first agent in
about 40 lines.
A working voice receptionist: hooked to your calendar, branded, deployed. Walk through the codec tutorial in under 10 minutes.
# Voice, chat and any MCP client can call it.
SKILL_NAME = "weather"
SKILL_DESCRIPTION = "current conditions for a city"
SKILL_TRIGGERS = ["weather", "forecast"]
SKILL_MCP_EXPOSE = True
def run(task, context=""):
city = extract_city(task)
return forecast(city)
Come build with us
on Codec.
Drop a star, open an issue, ship a skill. The framework is free forever. The agency around it is how we eat.
Say it. Watch it run.
Pick a voice command and Codec plays its real loop — transcribe, act, confirm — at believable speed, in the framework’s own monospace.
One voice line, three steps.
Codec hears the words, picks the intent, routes the tools, and reports back. Pick a command — it runs exactly as a deployed agent would.
Order online — start this week
Secure Stripe checkout. Confirmation in minutes, kickoff within 24 hours.