Skip to content

v2.md

Latest commit

 

History

History
50 lines (36 loc) · 3.44 KB

v2.md

File metadata and controls

50 lines (36 loc) · 3.44 KB

V2 Master Plan: Anahata AI Framework

This document is the single source of truth for the V2 refactoring of the anahata-ai framework (formerly gemini-java-client). It consolidates the strategic vision, architectural plans, UI/UX goals, and technical debt from all project documentation.

1. Unified Vision & The Flywheel Strategy

The project follows a "flywheel" business model:

  • The "Engine" (anahata-ai): This is the core commercial product, a pure-Java, enterprise-grade AI framework. It is monetized via a dual AGPLv3/Commercial license.
  • The "Car" (anahata-netbeans-ai): This is the free, open-source flagship application that showcases the engine's power and drives adoption.

Our key competitive differentiator is deep, interactive, and visual integration into desktop/IDE environments.

2. V2 Core Architectural Pillars

2.1. Model Agnosticism & Project Modularity

The primary technical goal is to decouple the framework from any specific AI provider.

  • Action: Split the project into three modules: anahata-ai-core, anahata-ai-gemini, and anahata-ai-swing.
  • Action: Implement a model-agnostic domain in anahata-ai-core for all conversation elements.
  • Action: Create a ContentProducer interface to abstract away all provider-specific object creation.

2.2. The Active Workspace Model

This is the new paradigm for state management.

  • Concept: A client-side ActiveWorkspaceManager will hold the definitive state of all StatefulResources.
  • Action: All tool calls will become ephemeral and aggressively pruned.
  • Action: Before each API call, the state from the ActiveWorkspaceManager will be injected into the prompt.

2.3. Instance-Based "Servlet-Style" Tooling

  • Concept: Migrate from static tool methods to an instance-based model.
  • Action: Create a BaseTool class that tools can extend to get access to the current Chat instance and the ActiveWorkspaceManager.

2.4. Advanced Schema Generation

  • Goal: Replace the custom GeminiSchemaGenerator with a mature, industry-standard library like jackson-module-jsonSchema.
  • Benefit: This will provide full, compliant JSON Schema generation and make the framework easily adaptable to other LLMs.

2.5. Hierarchical Chat Management

  • Goal: A Chat instance will have the ability to spawn and manage child Chat instances for complex, parallel workflows.

3. UI/UX Refactoring Plan

3.1. In-Chat Interactive Tool Prompter

  • Action: Replace the disruptive modal dialog with interactive components rendered directly within the chat stream.
  • Features: Each component will have its own controls for individual approval, denial, and setting session-level preferences (Always/Never).

3.2. Anahata Session Navigator

  • Action: Create a new AnahataInstancesTopComponent that displays a live list of all active chat sessions with real-time status.

4. LocalFiles Tool Evolution

  • Goal: Evolve the listDirectory tool to return a rich, polymorphic List<FileSystemEntry> instead of a simple List<String>.
  • Benefit: This provides structured, unambiguous metadata about files, directories, and symlinks, eliminating the need for inefficient and fragile shell calls (ls -ld) to determine entry types.
  • Action: Implement a new findFiles(path, recursive, glob) method to provide a structured, superior alternative to shell find and grep commands.