@thelapyae/geniclaw 1.1.11 → 1.2.2

package/README.md

@@ -1,14 +1,22 @@
 # 🦁 GeniClaw - Your AI Terminal Assistant
 
-**GeniClaw** is a powerful, local-first AI assistant manager that connects **Google Gemini** to your workflow. It runs as a background service (daemon) on your machine, managing conversation history, project sessions, and token usage efficiently.
-
-## 🚀 Features
-
--   **🧠 Multi-Model Support**: Switch between `gemini-2.5-flash`, `pro`, and other models instantly.
--   **📂 Session Management**: Create separate "folders" (sessions) for different projects or topics. Each session has its own isolated memory.
--   **💾 Smart Memory**: Automatically manages conversation context. It keeps recent history while pruning older messages to stay within token limits (default ~100k tokens), enabling long-term usage without crashing.
--   **🖥️ Terminal UI (TUI)**: An interactive command-line interface to manage everything.
--   **⚡ Queue-Based**: Robust message processing ensures no data is lost, even if you are offline.
+**GeniClaw** is a powerful, local-first AI assistant manager that connects multiple AI providers (Gemini, OpenAI, Claude, Groq, DeepSeek, Grok) to your workflow. It runs as a background service (daemon) on your machine, managing conversation history, project sessions, and token usage efficiently.
+
+## 🚀 Key Features
+
+-   **🧠 Multi-Provider Support**: 
+    -   **Google Gemini** (1.5 Flash/Pro, 2.0 Flash)
+    -   **OpenAI** (GPT-4o, GPT-3.5)
+    -   **Anthropic** (Claude 3.5 Sonnet, Opus)
+    -   **Groq** (Llama 3, Mixtral - Ultra Fast)
+    -   **DeepSeek** (Coder, Chat)
+    -   **xAI** (Grok Beta)
+-   **🔎 Web Search**: Integrated **Brave Search** allows the AI to Google things for you in real-time.
+-   **🔀 Task Routing**: Assign different models to different tasks! (e.g., Use Groq for fast Chat, Gemini for heartbeat/cron jobs).
+-   **📂 Session Management**: Create separate "folders" (sessions) for different projects.
+-   **💾 Smart Memory**: Automatically manages conversation context and prunes old messages to stay within token limits.
+-   **🖥️ Terminal UI (TUI)**: A beautiful interactive interface to manage keys, models, and sessions.
+-   **⚡ Queue-Based**: Robust message processing ensures no data is lost.
 
 ---
 
@@ -17,20 +25,23 @@
 ### Prerequisites
 
 -   **Node.js** (v18 or higher)
--   **Google Gemini API Key**: Get one for free at [aistudio.google.com](https://aistudio.google.com/).
--   **Telegram Bot Token** (Optional, for chatting via Telegram): Get one from [@BotFather](https://t.me/botfather).
+-   **API Keys**: You need at least one API key from a supported provider:
+    -   [Google AI Studio](https://aistudio.google.com/)
+    -   [OpenAI](https://platform.openai.com/)
+    -   [Anthropic](https://console.anthropic.com/)
+    -   [Groq](https://console.groq.com/)
+    -   [DeepSeek](https://platform.deepseek.com/)
+    -   [Brave Search](https://brave.com/search/api/) (Optional, for Web Search)
+-   **Telegram Bot Token** (Optional, for chatting via Telegram).
 
 ### Installation
 
 1.  **Install via NPM**:
-    ```bash
     ```bash
     npm install -g @thelapyae/geniclaw
     ```
-    *Note: The `-g` flag installs it globally so you can run `geniclaw` from anywhere.*
 
 2.  **Run Setup**:
-    This will guide you through entering your API keys.
     ```bash
     geniclaw setup
     ```
@@ -42,87 +53,52 @@
 GeniClaw is managed entirely through its **Interactive TUI**.
 
 ### 1. Start the TUI
-Run the following command in your terminal:
 ```bash
-./bin/geniclaw.js
-# Or if you linked it:
 geniclaw
 ```
 
 ### 2. Main Menu Options
--   **Start Daemon**: Launches the background process (Queue Processor & Telegram Bot).
--   **Stop Daemon**: Stops all background processes.
--   **Check Status**: Shows if the daemon is running and recent activity.
--   **📂 Manage Sessions**: Create or switch between different conversation contexts.
--   **🧠 Change Gemini Model**: Select the AI model you want to use.
--   **View Logs**: Inspect what's happening under the hood.
+-   **🤖 Select AI Provider & Model**: Choose which brain you want to use (Gemini, GPT-4, Claude, etc.).
+-   **🔑 Manage API Keys**: Enter keys for different providers.
+-   **🔀 Task Routing**: Configure which provider handles Chat vs. Background Tasks.
+-   **🔎 Web Search Config**: Set your Brave Search API key to enable internet access.
+-   **Start/Stop Daemon**: Control the background process.
 
 ### 3. Chatting
-Once the daemon is running (`Start Daemon`), you can interact with it via:
--   **Telegram**: If configured, just chat with your bot.
--   **CLI (Manual)**: You can send a one-off message from the terminal:
-    ```bash
-    ./geniclaw.sh send "Hello Gemini, how are you?"
-    ```
-
----
-
-## 📂 Session Management (Folders)
-
-GeniClaw allows you to organize multiple conversations.
-
-1.  Open the TUI: `geniclaw`
-2.  Select **Manage Sessions**.
-3.  **Create New Session**: Name it (e.g., `coding-project`, `personal-notes`).
-4.  **Switch Session**: Select it to make it active.
-
-**How it works**:
--   When you switch to `coding-project`, GeniClaw creates a folder at `~/.geniclaw/sessions/coding-project/`.
--   All history and memory for that conversation are stored there.
--   Switching back to `default` restores your previous context instantly.
-
----
-
-## ⚙️ Memory & Token Management
-
-To support **Long Term Usage**, GeniClaw monitors your token usage.
+-   **Telegram**: Chat with your bot.
+-   **CLI**: `geniclaw send "Hello!"`
 
--   **Automatic Pruning**: If a conversation exceeds the safe limit (default ~100k tokens), GeniClaw removes the oldest messages to make room for new ones. This prevents "Context Length Exceeded" errors.
--   **Check Usage**:
-    -   Type `/stats` in the chat to see current token usage.
--   **Reset**:
-    -   Type `/reset` or use the **Reset** command in TUI to clear the current session's memory.
+### web Search
+Just ask:
+-   "Search for the latest heavy metal news"
+-   "Google the price of Bitcoin"
+-   "/search Who won the super bowl?"
 
 ---
 
-## 🏗️ Architecture & Storage
+## 🔒 Safety & Technology
 
-### Where are my files?
-By default, GeniClaw stores everything locally in your home directory:
-`~/.geniclaw/`
+GeniClaw creates a secure, local-first environment for your AI interactions.
 
--   `config.json`: Your settings (API keys, active session).
--   `sessions/`: Folders containing history files for each session.
--   `queue/`: Temporary files for messages being processed.
--   `logs/`: Log files for debugging.
+### Technology Stack
+-   **Runtime**: Node.js & TypeScript.
+-   **State Management**: Local JSON files (No external database required).
+-   **API Integration**: Uses official SDKs (Google Generative AI) and standard REST APIs (OpenAI-compatible) via `node-fetch`.
+-   **Background Service**: A lightweight daemon process manages the message queue.
 
-### How it works
-1.  **Input**: You send a message (Telegram/CLI).
-2.  **Queue**: The message is written to `queue/incoming/`.
-3.  **Processor**: The `queue-processor` picks it up.
-    -   It loads the **Active Session** history.
-    -   It checks **Token Limits**.
-    -   It sends context + message to **Gemini API**.
-4.  **Response**: Gemini's reply is saved to history and written to `queue/outgoing/`.
-5.  **Output**: The interface (Telegram/CLI) sends the reply back to you.
+### Why is it Safe?
+1.  **Local Data**: All conversation history, sessions, and logs are stored **locally on your machine** in `~/.geniclaw/`. Your data never leaves your computer except to go directly to the AI provider.
+2.  **Direct Connection**: GeniClaw connects **directly** to the AI APIs (Google, OpenAI, etc.). There is no "middleman" server collecting your data.
+3.  **Open Source**: The code is transparent. You can inspect exactly how your keys and data are used.
+4.  **Key Security**: API keys are stored in a local config file (`config.json`) on your machine.
 
 ---
 
-## ⚠️ Things to Take Care Of
+## ⚠️ Important Notes
 
-1.  **API Costs**: While many Gemini models have a free tier, heavy usage (especially with large contexts) might incur costs if you are on a paid plan. Use `/stats` to monitor usage.
-2.  **Daemon**: The background process must be running to reply. If it stops, just run **Start Daemon** again.
-3.  **Privacy**: Your history is stored **locally** in plain text JSON files in `~/.geniclaw`. Keep your computer secure.
+1.  **API Costs**: Using commercial APIs (OpenAI, Anthropic) is not free. Costs depend on your usage. Groq and Gemini (Free Tier) offer free options.
+2.  **Privacy**: While GeniClaw is local, the AI Providers (Google, OpenAI, etc.) have their own data policies.
+3.  **Daemon**: The background process must be running for GeniClaw to reply.
 
 ---