The AirMouse AI platform is open to builders. REST API, native SDKs, CLI tools, webhooks, and first-class AI model integrations — everything you need to automate, extend, and ship faster.
Full CRUD control over workflows, devices, clipboard, voice sessions, and AI inference — all over HTTPS with JSON. OpenAPI 3.1 spec included.
# Trigger a workflow via the REST API curl -X POST https://api.airmouse.ai/v1/workflow/run \ -H "Authorization: Bearer am_sk_live_..." \ -H "Content-Type: application/json" \ -d '{ "workflow_id": "wf_standup_9am", "trigger": "api", "params": { "user_id": "usr_01J8XK", "device_id": "dev_macbook_pro", "dry_run": false } }' # Response { "id": "run_01J8XKABCDEF", "status": "running", "workflow_id": "wf_standup_9am", "steps_total": 4, "steps_done": 0, "started_at": "2025-01-14T09:00:01Z", "estimated_ms": 1240 }
Every SDK ships with full type safety, async/await support, automatic retries, and built-in webhook verification.
Full TypeScript generics, Node.js and browser compatible. Works with Next.js, Deno, and Bun out of the box.
npm install @airmouse/sdk
asyncio-native with Pydantic models. Integrates cleanly with LangChain, FastAPI, and Jupyter notebooks.
pip install airmouse-sdk
Idiomatic Go with context propagation, structured errors, and zero external dependencies. Perfect for server-side automation.
go get github.com/airmouse/sdk-go
Swift Concurrency–ready, SwiftUI bindings included. Build iOS Shortcuts extensions that talk directly to your workflows.
// Package.swift .package(url: "https://github.com/airmouse/sdk-swift")
Coroutines-first design, Flow support, and Jetpack Compose–compatible. Build Android companion apps with full API coverage.
// build.gradle.kts implementation("ai.airmouse:sdk-kotlin:0.7.1")
Zero-cost async with Tokio, fully typed with serde. Ideal for building high-performance automation agents or system-level tools.
# Cargo.toml airmouse = "0.3.0"
import { AirMouse } from '@airmouse/sdk'; const am = new AirMouse({ apiKey: process.env.AM_API_KEY }); // Define a multi-step workflow in TypeScript const workflow = am.workflow.define({ name: 'Deep Work Session', trigger: { type: 'voice', phrase: 'focus mode on' }, steps: [ { action: 'app.launch', params: { app: 'VS Code' } }, { action: 'notifications.mute', params: { duration: 90 } }, { action: 'spotify.play', params: { playlist: 'Deep Focus' } }, { action: 'ai.remind', params: { after: 90, message: 'Time for a break!' } }, ], }); await workflow.save(); console.log(`Workflow saved: ${workflow.id}`);
The SDK is designed to feel like writing plain TypeScript — not gluing API calls together. Define workflows, compose steps, and listen to events with a clean, typed interface.
airmouse emulateAuthenticate with a long-lived API key for server-side code, or short-lived JWTs for client-side flows. Scoped permissions, key rotation, and audit logging built in.
# Long-lived API key (server-to-server) const client = new AirMouse({ apiKey: 'am_sk_live_TpGkY8...', // keep secret! }); ───────────────────────────────────────── # Short-lived JWT (client-side, scoped) const token = await am.auth.createToken({ subject: 'usr_01J8XK', scopes: ['workflows:read', 'devices:write'], expiresIn: '15m', }); # Use in client SDK const userClient = new AirMouse({ token }); await userClient.workflow.list(); // ✓ await userClient.auth.deleteKey(); // ✗ not in scope
Register HTTPS endpoints and receive signed payloads for every platform event. Replay, retry, and inspect every delivery in the dashboard.
Fired when a device pairs, goes offline, reconnects, or changes capability state.
Track every workflow run from trigger through each step to completion — or failure with full stack trace.
Receive transcriptions, intent classification results, and model inference completions as they stream.
AirMouse ships with native adapters for the leading AI providers. Swap models in a single config change — no code rewrites.
GPT-4o, GPT-4 Turbo, and o1 — voice transcription via Whisper, image via Vision.
Claude 3.5 Sonnet and Haiku — ideal for complex workflow reasoning and long-context tasks.
Run any open-source model — Llama 3, Mistral, Stable Diffusion — in your workflows instantly.
Bring your own GGUF, ONNX, or HTTP-based model. The adapter interface is fully documented.
Deploy workflows, manage API keys, stream logs, run the local emulator, and scaffold new integrations — all from your terminal.
airmouse deploy packages and pushes your workflow bundle# Install the AirMouse CLI npm install -g @airmouse/cli # Authenticate with your account airmouse login ✓ Logged in as shafqat@company.com # Run a workflow locally for testing airmouse run workflows/standup.ts --device emulator ▶ step 1/4 — app.launch VS Code ✓ 210ms ▶ step 2/4 — notifications.mute ✓ 12ms ▶ step 3/4 — spotify.play ✓ 381ms ▶ step 4/4 — ai.remind (scheduled) ✓ queued ✔ Workflow completed in 604ms # Deploy to production airmouse deploy --env production ✔ Deployed: wf_standup_9am@v3
Core protocol layer, CLI, all 6 SDKs, and the local emulator are MIT-licensed on GitHub. We build in public.
Core infrastructure, all SDKs, and the protocol layer are MIT-licensed. Ship features, fix bugs, and shape the roadmap — as part of the community that uses AirMouse every day.
18,000+ builders, weekly office hours, and a Discord where the AirMouse core team hangs out daily.
9,000+ members. Real-time help, showcase channels, release previews, and direct access to the core team.
Join Discord →RFC proposals, architecture discussions, Q&A, and community showcases — searchable, indexed, and alive with activity.
View discussions →Every Thursday at 5 PM UTC — 45 minutes with the engineering team. Live demos, roadmap previews, and open Q&A.
Add to calendar →Fork, fix, PR — it really is that simple. Here's the four-step path from zero to merged.
Browse good first issue and help wanted labels on GitHub. Each has a spec and expected output attached.
Fork the repo, create a branch off main, and run pnpm dev to start the local dev loop with hot reload.
All contributions need passing unit and integration tests. Our CI runs on every commit — green build, green light.
We review within 48 hours. Maintainers leave detailed, constructive feedback. Typical time-to-merge: 4 days.
Free developer account. 10k API requests per month on the free tier. No credit card required.