MCP servers, developer tools, and browser extensions we built for our own work and released under MIT. Use them, fork them, contribute.
Token-based SwiftUI from the ground up.
All visual decisions live in design tokens. Swap a theme at runtime without touching component code.
Fully audited for strict concurrency — safe with async/await and Swift actors, no warnings suppressed.
Default — System semantic colors — fully adaptive, no preset required.
DesignFoundation has its own page with theme previews, installation options, and the full component reference.
Full DesignFoundation page401 tools covering the full GHL API v3 — contacts, conversations, pipelines, calendars, invoices, knowledge bases, ad publishing, social planner, and more. Talk to your CRM in natural language.
SwiftUI + WKWebView browser components for macOS agent applications — a full multi-tab browser panel, pinned-site sidebar, bookmarks, JS console capture, and screenshot tooling.
21 tools for scheduling and publishing across 10 social platforms — Instagram, YouTube, TikTok, LinkedIn, X, Bluesky, and more — via natural language.
Codebase intelligence engine that builds a knowledge graph via tree-sitter AST parsing and exposes it as an MCP server. 23–40% fewer tool calls on complex tasks.
47 tools covering the full LMEX exchange API — spot trading, futures, wallet management, market data, and order types including OCO and PEG.
Claude drafts reply options as you browse Reddit threads. You review, edit, and post manually. Nothing is ever submitted automatically.
Production-grade SwiftUI design system with token-based theming, style-swappable components, Liquid Glass support, and Swift 6 concurrency safety.
A unified chat interface across cloud and on-device AI providers. OpenAI, Anthropic, and Apple Intelligence share one protocol, message model, and optional SwiftUI components — swap providers with a one-line change.
Adds on-device Apple MLX inference to any AIChatKit app. Models download from Hugging Face once, then run entirely offline on Metal GPU and the Neural Engine — text and vision-language models both supported.
Adds on-device GGUF inference via llama.cpp to any AIChatKit app. Any GGUF model runs in-process on Metal GPU, with no network calls after the initial model download.
An embeddable, on-device long-term memory engine for AI chat apps — a bi-temporal knowledge graph with hybrid BM25 + embedding retrieval and LLM-driven fact extraction and correction.
A fully customizable Kanban board for SwiftUI on iOS, iPadOS, and macOS — cross-column pointer drag with autoscroll, a full keyboard move-mode, and VoiceOver parity for every drag interaction.
On-device voice dictation, global hotkey command routing, and AI-assisted text rewriting for macOS. Local ASR via Apple Speech or downloadable Whisper/Parakeet models — no cloud dependency required.
A full-source, production-quality on-device AI chat app for macOS — MLX inference, AiVoiceKit dictation, AiPersona memory, and a UI assembled from a real design system. Requires a DesignFoundationPro license to build.
We build MCP servers and AI agents as part of our consulting work. If your team needs a tool like these for an API we haven't covered yet, get in touch.