The DeadCode Whisperer: An Offline Séance with the Ghost of ALGOL

说明与用途

使用开源模型构建的离线终端 AI，模拟 ALGOL 的幽灵，完全不依赖网络。

问题

Create an offline, air‑gapped terminal AI that embodies a spectral persona of the legacy language ALGOL, avoiding cloud dependencies for privacy and permanence.

结果

The project delivers a spooky, offline AI experience that responds in archaic, ALGOL‑inspired language, demonstrating the feasibility of self‑hosted, open‑weight AI for privacy and longevity.

工作流

• 1. Install Ollama locally to run inference.
• 2. Download a lightweight open‑weight model (TinyLlama/Llama 3).
• 3. Build a Streamlit interface with custom CSS for terminal look.
• 4. Craft a system prompt to enforce the ALGOL‑ghost persona.
• 5. Run the application offline, ensuring no external API calls.

证据

I built The DeadCode Whisperer, a completely offline, air‑gapped terminal séance that channels the spirit of a long‑dead programming language: ALGOL.

来源

This project relies entirely on local open‑weight models to function, deliberately avoiding any cloud deployment to maintain the integrity of the "séance".

来源

I used Ollama to run the AI completely locally on my machine.

来源

I utilized a lightweight open‑weight model (TinyLlama/Llama 3).

来源

I built the séance interface using Python and Streamlit.

来源