Skip to content

Releases: guojun21/thalamus

v0.2.0 — Desktop App Launcher

07 Mar 15:16
@guojun21 guojun21
v0.2.0
a29de6c

Choose a tag to compare

View all tags
v0.2.0 — Desktop App Launcher Latest
Latest

What's New

Desktop App Launcher (macOS)

Native macOS .app that provides a one-click experience for running Thalamus:

  • One-click start — double-click to launch, auto-starts thalamus-py backend
  • Login UI — built-in Cursor PKCE login with auto token save
  • API Test Panel — test models list and send messages with model selection
  • Native macOS — built with Swift + WKWebView, no Electron, 269KB

Download & Install

  1. Download Thalamus-macOS.zip below
  2. Unzip and drag Thalamus.app to /Applications
  3. Double-click to launch

Note: Requires Python 3.10+ with thalamus-py dependencies installed. Run bash desktop-app/build.sh to build from source.

Build from Source

cd thalamus
python3 -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
cd desktop-app && bash build.sh
Assets 3
Loading

v0.1.0 — Initial Release

07 Mar 12:17
@guojun21 guojun21
v0.1.0
b4ba669

Choose a tag to compare

View all tags
v0.1.0 — Initial Release

Thalamus v0.1.0 — Not the mind. The gateway to it.

Smart local proxy that lets Claude Code transparently use Cursor's models with enhanced tool calling, auto-continuation, and intelligent model fallback.

Highlights

🔮 Lazy Tool Loading Protocol (LTLP)

Claude Code sends 40+ tool definitions (~27K tokens). Thalamus compresses them into ultra-compact stubs (~1KB) and loads full schemas on-demand. The model learns correct parameters through few-shot context — zero wasted tokens.

🔄 Auto-Continuation with task_complete

Cursor's API provides no native stop_reason. When the model outputs text without tool calls, Thalamus automatically retries and merges text + tool calls into a single response. Fully transparent to Claude Code.

🛡️ Smart Model Fallback

First-token timeout detection with configurable retry chains across multiple models.

📡 Dual API

Serves both /v1/messages (Anthropic) and /v1/chat/completions (OpenAI) from the same Cursor backend.

Quick Start

git clone https://github.com/guojun21/thalamus.git
cd thalamus
python3 -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env   # Set your CURSOR_TOKEN
python server.py

# Connect Claude Code
export ANTHROPIC_BASE_URL=http://localhost:3013
export ANTHROPIC_API_KEY=thalamus-proxy
claude

Tech Stack

Python 3.10+ · FastAPI · httpx/h2 · protobuf · Pydantic

Loading