Skip to content

Prompt.md

Latest commit

 

History

History
228 lines (127 loc) · 2.85 KB

Prompt.md

File metadata and controls

228 lines (127 loc) · 2.85 KB
Raw

Phase 9 Prompt

Plugin Framework

Create the architecture that allows NAME to become extensible.

Design

Plugin Manifest

Plugin Permissions

Plugin Lifecycle

Plugin Metadata

Plugin Events

Plugin Settings

Plugin APIs

Plugin Installation

Plugin Updates

Plugin Registry

Plugin Validation

Plugin Loader interfaces

Future plugins must be able to add

Pages

Widgets

Services

Settings

Commands

Agents

Tools

without modifying existing code.

No implementation.

Architecture only.

Phase 10 Prompt

Linux Runtime Interfaces

Prepare the abstraction layer for future Linux integration.

Create service interfaces for

Filesystem

Shell

PTY

Permissions

Environment Variables

Package Manager

Python Runtime

llama.cpp Runtime

GPU Detection

Hardware Detection

Storage

Process Management

Do not execute commands.

Create only interfaces and documented stub methods.

Flutter UI must never communicate directly with Linux.

Everything must pass through service abstractions.

Phase 11 Prompt

Command Palette

Build a professional Command Palette similar to VS Code.

Support mock functionality for

Page Navigation

Recent Files

Recent Models

Recent Chats

Agent Launch

Workspace Commands

Quick Settings

Search

Keyboard Navigation

Command Categories

Future plugins must be able to register commands dynamically.

Mock implementation only.

Phase 12 Prompt

Workspace UI

Transform the application into a true AI Operating Environment.

Implement

Resizable Panels

Tabbed Views

Collapsible Sidebars

Bottom Console

Status Bar

Notification Center

Activity Bar

Workspace Layout Manager

Everything should use placeholder content.

No backend.

Phase 13 Prompt

Architecture Hardening

Perform a complete architecture review.

Identify

Code smells

SOLID violations

Dependency issues

Widget rebuild issues

Performance bottlenecks

Naming inconsistencies

Folder inconsistencies

Missing documentation

Unused providers

Dead code

Missing abstractions

Refactor the project to enterprise quality while preserving behavior.

Produce a detailed architecture report describing every improvement.

Overall recommendation

Instead of continuing with ad hoc prompts, maintain a development roadmap with milestones:

  1. Foundation (completed)
  2. Architecture audit
  3. Design system
  4. Navigation framework
  5. Workspace/window management
  6. Model management
  7. Agent framework
  8. Memory framework
  9. Plugin system
  10. Linux runtime abstractions
  11. Command palette
  12. Workspace UI
  13. AI runtime integration (later)
  14. Python runtime (later)
  15. llama.cpp integration (later)
  16. Autonomous agent execution (later)

This incremental approach is much more likely to produce a maintainable, enterprise-grade codebase than repeatedly asking the AI to generate large sections of the project in one pass.