Hi, I'm Amen 👋

Building voice AI, agent tooling, and developer infrastructure.

🌐 amenophis.dev

X · LinkedIn · Threads · Instagram · Facebook

Stack

Languages: Go · TypeScript · Rust · Python
Domains: voice AI (STT/TTS/LLM streaming) · AI agents & MCP · developer tooling
Infra: Cloudflare Workers · Docker · GitHub Actions · macOS/Linux

Featured

• Shot Crisp Swift — macOS screenshot studio — capture, annotate, OCR, pin, share; menu-bar native with an MCP bridge for agents.
• Vocito Swift — Dictation that types for you — push-to-talk voice typing for macOS. Talk naturally, false starts and all; Vocito types the sentence you meant, formatted for wherever you're writing. Private beta.
• My Notes — Side Panel Notepad TypeScript — Local-first, Notion-like notes in the Chrome side panel — block editor, revision history, optional Drive/Dropbox sync.
• cadence Go — Provider-agnostic Go engine for low-latency STT→LLM→TTS voice pipelines — streaming engines behind uniform interfaces, swap vendors with a config change. Owns the hot path.
• xPhone (Go version) Go — The embedded telephony data plane for Go. Place and answer calls on any SIP trunk — or be the trunk — with carrier failover, full SIP-signaling visibility, and call audio delivered as PCM frames on a channel, ready for STT or a voice agent. Battle-tested across 500,000+ production calls.
• Vonzio TypeScript — The runtime for production agents — bring your own model. Open-source, self-hostable, isolated Docker workspaces per session.
• Foreman TypeScript — Runs software missions unattended and shows you everything when you're back. A director plans, delegates to workers, and verifies; any provider, honest cost, a default for every ask, your phone in the loop.
• elefante TypeScript — The open, Git-native memory protocol for MCP agents: portable, inspectable, versioned, and human-editable.
• claude-voice TypeScript — Give Claude Code a voice — selective, tasteful spoken summaries and attention cues. Quiet by default.

Also building

• xbridge Rust — Phone calls for every other language: one self-hosted binary turns SIP calls into WebSocket audio + REST call control. Twilio-compatible framing, so existing voice apps port over with the URL swapped.
• xPhone (Rust version) Rust — The same phone line for Rust — SIP and RTP implemented from scratch, zero telephony crate dependencies. Call audio as crossbeam channels of PCM frames; powers xbridge in production.
• fakepbx Go — In-process SIP server for Go tests. Real SIP over loopback — no Docker, no Asterisk, no hardcoded ports.
• xpbx Go — A real PBX in one docker run — Asterisk with a web UI for extensions, trunks and dialplans. The five-minute test bench for anything you build with xphone or xbridge.
• oxiclean Rust — macOS storage cleaner — reclaims disk from package-manager caches, Docker artifacts, and orphaned app data. brew install getoxiclean/tap/oxiclean
• Trokky TypeScript — Modern, composable CMS for developers — in private beta, with an open-source CLI for backup, restore, export and import.
• gh-setup Go — Declaratively configure GitHub accounts, repos, and settings from a YAML file
• tts-bridge Rust — TTS APIs speak HTTP. Voice apps speak WebSocket. This bridges the gap.
• shopcart TypeScript — Save Amazon products to custom lists. Chrome extension for organizing your wishlist

Writing

• The anatomy of time-to-first-audio in a voice agent — I measured my voice pipeline's wait — sentence buffer, VAD endpointing, and a live LLM→TTS chain. Post-commit, my code is ~14% of it; adding endpointing roughly doubles the share I control.
• Giving AI agents a real phone line — A browser voice demo proves your model can talk. It doesn't prove your stack can survive a phone network — so I measured mine: frame-timing distributions at 50 concurrent calls, plus a live-trunk check.

---

_{This README is generated from a manifest — updated when I ship, not when I remember.}