Wave Terminal AI History & Context Toolkit – Real-Time Chat Watcher, Auto-Archiver & Daily Context Handove

[adamdziewit-eng]

Hi everyone! 👋

I built a complete AI History & Context Toolkit for Wave Terminal (Windows & Linux) that makes working with Wave AI and local Ollama models much more productive. With this setup, you never lose a conversation, code snippet, or project context across sessions.

✨ What it does:

1. Real-Time Background Wave AI Watcher (save_online_history.py)

   • Silently monitors Wave AI sidebar chats via Wave's local RPC socket (wave.sock) and SQLite metadata.
   • Automatically saves all conversations in reverse chronological order (newest sessions and newest exchanges always at the top!) into:
     • Markdown (online_history.md) – ready for 1-click viewing in Wave's Preview widget.
     • Interactive HTML (online_history.html) – dark-mode dashboard with live search filtering and 1-click "Copy Code" buttons.
     • Plain Text (online_history.txt).

2. Smart Auto-Archiving (archive/)

   • Keeps active history files fast and lightweight (~1,500 lines max), automatically moving older User -> Model exchanges into ai_history/archive/ (.md, .html, .txt) without breaking code blocks.

3. 1-Click Daily Context Handover Pill (Generate AI Handover Widget)

   • Generates a compact context_YYYY-MM-DD_HH-MM.md summary of your day's work (touched files, chronological prompt list, and detailed summary of the last 5 exchanges).
   • Attach this single file to a fresh AI chat the next day so the model immediately knows where you left off!

4. Local Ollama Terminal Client (run-ai.ps1) & Cross-Platform Linux Migrator (wave_migrator.py)

   • Includes a terminal UI for local Ollama models with ANSI syntax highlighting, slash commands (/load, /save, /search, /copy, /history), and a 1-click Windows ↔ Linux config migrator.

🔗 GitHub Repository & 1-Click Installer:

👉 https://github.com/adamdziewit-eng/waveterm-ai-toolkit.git

Feedback and suggestions are very welcome!