Terminalis AGI - Ultimate Local AI Ecosystem

Terminalis AGI is a comprehensive, locally runnable AI ecosystem designed for power users and developers. It provides a suite of tools, numerous AI models, and a flexible agent orchestration system, all intended to run on your local machine.

Key Features

• Local First: Designed to run primarily on your local hardware.
• Ollama Integration: Seamlessly manages Ollama for running various open-source Large Language Models.
• Multiple Pre-configured AI Agents: Includes specialized agents for tasks like advanced reasoning, code generation, data analysis, web research, and more.
  • ImageForge Agent: Text-to-image generation using advanced diffusion models.
  • Video Processing Utilities: Get video info, extract frames, and convert video segments to GIF.
  • AudioMaestro Agent: Audio file analysis (get info), format conversion, and Text-to-Speech (TTS).
  • Enhanced CodeMaster: Experimental AI-assisted code modification in projects, code explanation, and generation of new code modules/classes.
• Sophisticated Orchestration Engine:
  • MasterPlanner Agent (Evolving): Decomposes complex user requests into multi-step execution plans. Features include:
    • Step-level validation and robust retry mechanisms.
    • Basic parallel execution for independent tasks.
    • Targeted plan revision based on detailed failure context.
    • Preliminary Knowledge Base query to inform planning.
  • Contextual Agent Selection: Orchestrator utilizes UI context and zero-shot intent classification for improved agent routing.
  • Conversational Memory: Basic conversation history is maintained and provided as context to MasterPlanner.
• Persistent Knowledge Base (ChromaDB):
  • Stores and retrieves textual information (document excerpts, web summaries, code explanations/generations) via semantic search.
  • Includes a "Knowledge Base Explorer" in the Web UI for direct user querying.
  • Automatically augmented by agents like DocumentProcessor (via UI), WebCrawler, and CodeMaster.
• Extensible Agent Configuration: Agent capabilities and models can be configured via agents.json.
• Model Management: Model downloads are managed via models.conf.
• Web UI (Streamlit): Provides an interactive interface with various operation modes, including the new KB Explorer.
• Core Capabilities: Document processing, web intelligence, project scaffolding, system information.
• Version-Pinned Dependencies: Uses *_requirements.txt files with pinned versions for Python packages to ensure stability and reproducibility. All dependencies are pinned to specific versions.
• Automated Setup: A single bash script automates the installation of dependencies and components.

Installation

System Requirements

• A Linux or macOS environment.
• Python 3.8+ and Pip.
• Docker (for some potential future integrations, though primarily uses local Ollama).
• Standard build tools (git, curl, wget, build-essential, cmake, etc. - the installer attempts to manage these).
• Sufficient disk space (min. 100GB recommended for multiple models).

Running the Installer

1. Ensure you have the installer script (e.g., script_content.txt or a named .sh file).
2. Make it executable: chmod +x your_script_name.sh
3. Run the installer: bash your_script_name.sh

The main installation directory will be ~/.terminus-ai. All logs and generated files will be stored there.

Post-Installation

Launching Terminalis AGI

To launch the Terminalis AGI ecosystem:

cd ~/.terminus-ai
python3 launch_terminus.py

This will start the Ollama server (if not already running) and the Streamlit Web UI. You can access the UI at http://localhost:8501.

Basic Usage

The Web UI provides several operation modes:

• Multi-Agent Chat: Interact with multiple AI agents simultaneously. Can optionally use MasterPlanner for complex requests and benefits from contextual agent selection and conversational memory.
• Document Processing: Upload and analyze documents.
• Web Intelligence: Perform web searches and analyze results.
• Image Generation: Create images from text prompts.
• Video Processing: Utilities for video file information, frame extraction, and GIF conversion.
• Audio Processing: Tools for audio analysis, format conversion, and Text-to-Speech.
• Code Generation: Scaffold new projects, modify existing code (experimental), explain code snippets, and generate code modules. (Explanations & generated modules are now stored in the Knowledge Base).
• Knowledge Base Explorer: Directly query and explore the contents of the system's persistent knowledge base.
• Other modes for data analysis, system information, etc., may be available.

Select agents, adjust parameters like temperature, and input your queries or tasks through the UI.

Configuration

• Models (models.conf): Located in ~/.terminus-ai/models.conf. This file defines which Ollama models are available for download and allows you to specify core models or a full set.
• Agents (agents.json): Located in ~/.terminus-ai/agents.json. This file configures the available AI agents, their models, specialties, and active status. You can customize this file to add new agents or modify existing ones.

Note on Dependencies

The project uses version-pinned Python dependencies listed in core_requirements.txt, frameworks_requirements.txt, utils_requirements.txt, and dev_requirements.txt. All dependencies are pinned to specific, stable versions to ensure stability and reproducibility of the environment.