Open‑Source Text‑to‑3D Modeling, Free Social Media Management, 60% Token Savings

CADAM – Text‑to‑3D Modeling

Traditional CAD tools such as SolidWorks or Fusion 360 require learning sketching, extrusion, and rotation, which is costly for non‑engineers. CADAM is an open‑source web application that accepts a natural‑language description, uses an AI to generate OpenSCAD code, and renders the model in the browser via WebAssembly, eliminating the need for local installation.

Verified generated models include a turbofan engine, turbine blades, a bevel‑gear transmission, a honeycomb bracket, and a threaded bottle. The tool is positioned for prototyping and concept validation rather than high‑precision industrial design.

git clone https://github.com/Adam-CAD/CADAM.git
cd CADAM
npm install
npm run dev

Repository: https://github.com/Adam-CAD/CADAM

BrightBean Studio – Open‑Source Social Media Management

Managing multiple social accounts (WeChat, Weibo, Xiaohongshu, LinkedIn, TikTok, etc.) typically requires separate logins and posting workflows. Commercial tools such as Buffer, Hootsuite, Sendible, or SocialPilot charge per account or per user.

BrightBean Studio is an open‑source platform that matches the feature set of Sendible and SocialPilot while imposing no limits on the number of accounts, workspaces, or team members.

Core capabilities:

Calendar‑drag scheduling for 10+ platforms (Facebook, Instagram, LinkedIn, TikTok, YouTube, etc.)

Rich‑text editor for content creation

Unified inbox aggregating comments, direct messages, and @mentions

Approval workflows, analytics dashboards, and media library

Option for self‑hosting (data under user control) or official hosted service

git clone https://github.com/brightbeanxyz/brightbean-studio.git
cd brightbean-studio
docker compose up -d

Repository: https://github.com/brightbeanxyz/brightbean-studio

pxpipe – Convert Text to Image to Reduce Token Costs

Most AI services charge by token, and the largest token consumer in a request is often static context such as system prompts, tool documentation, or conversation history.

pxpipe converts dense text into an image before sending it to the model. Because large models charge for images by pixel rather than by the amount of embedded text, a 1573×1248‑pixel image containing roughly 48 000 characters costs about 2 700 image tokens, compared with ~25 000 text tokens—a saving of roughly 90%.

The approach is suitable for scenarios where information density is high but line‑by‑line reading is unnecessary (e.g., system prompts, tool docs, summary histories). For content requiring precise textual understanding, plain text remains necessary.

pxpipe runs as a local proxy that intercepts traffic between Claude Code (or other APIs) and the model, operating transparently.

npm install -g pxpipe
pxpipe start

Repository: https://github.com/teamchong/pxpipe