v1.1.0

Clew v1.1.0 — Native AI Coding IDE

Local-first AI dev environment, zero telemetry, 18+ providers, MCP support.

Features:

• 🔒 Local code, zero telemetry
• 🧠 18+ providers: Claude, GPT, Gemini, DeepSeek, local models (Ollama, LM Studio)
• 🤖 Autonomous ReAct agents (plan → read → write → execute → observe)
• 🧩 Built-in MCP protocol support
• ⚡ Native Qt app, fast startup
• 📂 Smart, git-aware context management
• 💬 Real-time token/cost tracking
• 🎨 Many themes: Ember, Dark, Light, Espresso, Aurora, Graphite, Clay, Cursor, System — from warm dark palettes to bright, minimal looks

Install: prebuilt binaries for macOS and Windows (no Linux app yet), or build from source with Python 3.11/3.12 (git clone, venv, pip install -e ".[dev]", clew).

Security: AES-256-GCM encrypted API keys, sandboxed command execution, no third-party trackers.

Known issues: macOS icon cache glitch on first launch.

License: Apache 2.0 (commercial use allowed).

Links: github.com/clew-ide/clew · github.com/clew-ide/clew/issues · github.com/clew-ide/clew/discussions · github.com/ilyaosovskoi/Clew

v1.0.3

Clew v1.0.3 Release Notes

We are thrilled to announce the latest update to Clew! This version represents a massive leap forward, completely redefining the Clew experience. We've overhauled the visual design, introduced powerful new workflows, and expanded API support. It now looks and feels like an incredibly premium, high-end application.

✨ What's New

• Premium Visual Overhaul: The entire UI has been redesigned for a breathtakingly premium look and feel.
• Themes & Ambient Effects: You can now switch between Light, Dark, and System themes. We've also added stunning ambient background effects that mimic signals between synapses. Don't like them? You can easily disable them in the settings.
• Multi-Provider API Support: Clew now supports a wide range of cloud APIs, from Anthropic to z.ai and many others. Bring your API keys and switch between local and cloud models effortlessly.
• High-Speed API Server: To drastically improve speed and reduce latency, Clew now spins up a lightweight local server to handle all incoming and outgoing API requests.
• Integrated Code Viewer: When you select a project directory, a powerful Viewer opens on the right side. You can now read your entire codebase right there in the app without any hassle.
• Prompt Templates: We've added ready-to-use prompt templates directly in the workspace. Just pick one, insert your specific words, and hit run.
• Live Token Counting: You now have accurate, real-time token counting during agent runs and chats.
• Auto-Updater: Say goodbye to manual downloads! Clew now includes a built-in auto-updater.
• Detailed Local Model Settings: Fine-tune your local inference parameters directly in the UI.
• Bug Fixes: All known bugs and issues from previous versions have been completely squashed.

🛠 Future Sections (Coming Soon)

The workspace is being prepared for specialized agent modes. The General template is fully operational today, but keep an eye out for upcoming profiles like Office Worker and Heavy Code.

📦 Downloads

Below you will find the following assets attached to this release:

• Clew.dmg — The installer for macOS (Apple Silicon).
• Clew.exe — The Windows.
• Source code (zip) / Source code (tar.gz) — The complete source code for this release for developers.

Thank you for using Clew!

v1.0.0-beta

First public beta release of Clew.

Clew is a local-first AI IDE for macOS Apple Silicon that combines a code editor, autonomous AI agents and local LLM management in a single native application.

Highlights:

• 💻 Native macOS app (PySide6, no Electron)
• 🤖 Built-in AI agents
• 🧠 Local model manager with Hugging Face integration
• 🔒 100% local execution by default
• ⚡ MLX and llama.cpp support
• 📦 Simple .dmg installation (coming soon)

⚠️ This is an early beta release.

The core functionality is ready, but there are still rough edges. You may encounter bugs, indexing issues and UI inconsistencies.

This release is primarily intended to gather real-world feedback before the upcoming stabilization updates.

Feedback, bug reports and feature requests are highly appreciated.