Dunwell DevStudio — Development Roadmap
Select a milestone node on the timeline below to inspect live v1.0 features and upcoming architectural upgrades.
Foundational 3-Helper Local AI Workshop
Engineered to make local LLM software creation accessible to complete beginners and experienced developers alike, with zero cloud subscriptions and 100% local data ownership in %APPDATA%. View Dunwell DevStudio on the Microsoft Store →
Planner, Builder & Inspector Team
Automates software creation across three dedicated AI helpers: The Planner writes a clear checklist, The Builder saves a backup and writes the code, and The Inspector tests the program for errors.
Hardware & Bottleneck Detector
Automatically scans your GPU VRAM, CPU cores, and system RAM to place your PC into one of four Hardware Tiers and recommends the exact GGUF model sizes and quantisations for LM Studio.
1-Click Undo Backups & Vision Input
Automatically backs up project files before any edit for instant 1-Click Undo/Revert, supports Ctrl+V screenshot pasting for visual debugging, and includes voice microphone input.
Built-In Plain-English Guides
Includes a 30-Second AI Dictionary, a real-life Kitchen Countdown Timer walkthrough, and an interactive "Every Button Explained" reference right inside the app.
v1.1: Integrated "Advanced Mode" Toggle & 3D Game Engine Bridges
Scoped for the v1.1 update, Dunwell DevStudio introduces a Progressive Disclosure UI architecture controlled by an integrated Advanced Mode toggle switch—helping users progress from foundational AI coding to automated 3D engine pipelines without overwhelming beginners or bloating local LLM token budgets.
Core Multi-Agent Coding Workspace
Presents the streamlined 3-Helper interface, Quick Ideas, plain-English guides, and local model selection without intimidating engine configuration panels.
Zero External MCP Schema Bloat
Zero external Model Context Protocol (MCP) tool schemas are loaded into system prompts by default—keeping local models fast, lightweight, and responsive.
Core Workspace & Prompting Guides
Displays foundational documentation covering connecting LM Studio/Ollama, 3-Helper workflow basics, and hardware tier optimisation.
Automated C++ Gameplay & Details Panel Sliders
Agents write .h and .cpp files directly into your project's Source/ directory and run background headless compilation via Visual Studio C++ build tools to verify builds and self-correct errors. By inserting UPROPERTY(BlueprintReadWrite, EditAnywhere) macros, every gameplay variable is exposed as an interactive slider inside Unreal Engine's visual Details Panel.
Zero-Setup Spatial Maths & Batch Alignment
Communicates with Unreal Engine's built-in Python Editor Script Plugin over a local socket to fire temporary import unreal utility scripts—performing 3D mesh bounding-box calculations, level layout alignment, and batch asset processing in seconds before automatically cleaning up temporary files.
Unreal Engine 5.8, Blender & Godot Integration
Built around a local HTTP/SSE Model Context Protocol (MCP) client (e.g., http://127.0.0.1:8000/mcp) that connects cleanly to Unreal Engine 5.8, Blender, and Godot without requiring external Node.js or system PATH dependencies. Structured JSON tool menus prevent smaller local LLMs from hallucinating API functions.
3-Role Division, AGENTS.md & "Grill Me" Mode
Expands agent orchestration across a Lead Architect Agent, a C++ Gameplay Agent, and a Level & Editor Agent. Includes persistent AGENTS.md project rulebooks, automated Markdown context handoffs, and a native "Grill Me" Spec-Interview Toggle that interviews you on scope, camera style, and tunable parameters before writing code.
v1.2: Expanded Agent Templates & Workflow Automation
Planned for follow-up development after the v1.1 Engine Bridge release. Edit this section in roadmap.html whenever you are ready to announce specific v1.2 features.
Custom Multi-Agent Blueprints
Save, export, and share custom 3-Helper role prompts, test commands, and project starter templates across different machines.
Extended Local Toolchain Bridges
Additional local developer integrations and automated test-runner presets tailored to community feedback following the v1.1 release.