AI Unity Developer Assistant

Self-hosted automation tooling for Unity development workflows.

This project packages an internal automation runner called Claude Auto Daemon into a public-facing repository focused on portfolio quality, clean onboarding, and practical Unity team workflows. It combines a task queue, Git branch automation, scheduled QA analysis, a Flask web dashboard, Telegram controls, and optional local-model support through Ollama.

Why It Exists

Unity solo developers and small teams often juggle repetitive engineering chores:

• converting design notes into tracked implementation tasks
• running recurring code reviews and documentation audits
• managing integration branches for AI-assisted work
• checking project health when no active feature work is queued
• monitoring the automation remotely from a browser or Telegram

AI Unity Developer Assistant is built to automate those loops without pretending to be a Unity package. It is a Python service that orchestrates external tools around a Unity repository.

Core Features

Area	What it does
Task Queue	Stores work in tasks.json, prioritizes pending items, tracks status, retries, logs, and archive state
Git Automation	Creates task branches, manages an integration branch, collects diffs, merges, and optional pushes
Scheduled QA	Runs typed QA passes such as code review, architecture checks, documentation generation, tests, and performance audits
Web Dashboard	Exposes task lists, QA reports, logs, Git status, model settings, and runtime configuration through Flask
Telegram Bot	Supports status checks, task inspection, pause/resume, QA triggers, diffs, merges, logs, and token usage
Unity Project Mapping	Scans a target Unity repository, builds file maps, dependency graphs, and recent-change context for prompts
Local/Hybrid LLM Flow	Supports Claude Code CLI plus optional Ollama models for simple analyses and fallback scenarios
Workspace Isolation	Can clone a separate workspace for daemon activity so the main Unity repo stays free for manual work

Who This Is For

• solo Unity developers experimenting with AI-assisted implementation pipelines
• toolsmiths building internal automation around existing Unity repositories
• technical artists or producers who want visibility into queued work and QA output
• anyone looking for a showcase project that blends Python orchestration, Git workflows, and developer tooling

Architecture Overview

flowchart LR
    A["tasks.json / manual triggers"] --> B["Scheduler"]
    B --> C["TaskExecutor"]
    B --> D["QARunner"]
    C --> E["GitManager"]
    C --> F["Claude Code CLI"]
    D --> G["ProjectMap"]
    D --> H["Ollama / Claude provider"]
    B --> I["Flask Web UI"]
    B --> J["Telegram Bot"]
    E --> K["Unity project repository"]

Loading

The public repository name is AI Unity Developer Assistant, while many internal modules still use the older name Claude Auto Daemon. That naming mismatch is intentional for this release to preserve runtime compatibility.

Quick Start

1. Install Python 3.11+.
2. Create a virtual environment.
3. Install dependencies with pip install -r requirements.txt.
4. Copy config.example.json to config.json.
5. Optionally copy .env.example to .env.
6. Set general.project_path to your Unity Git repository.
7. Run python run.py --setup or edit config.json locally.
8. Start the service with python run.py.

Useful commands:

python run.py --help
python run.py --setup
python run.py --web-only
python run.py --daemon-only

Configuration at a Glance

The runtime reads from a local config.json file plus optional environment overrides.

• general: project path, daemon enablement, polling interval
• git: main/integration branches, task branch prefixes, push/merge behavior
• qa: scheduled QA runs, documentation depth, provider selection
• web: host, port, and optional dashboard authentication
• telegram: bot token, allowed users, notification toggles
• workspace: isolated clone settings for bot-driven work
• skills: external skills repository URL and local download path

Start from the shipped templates:

• config.example.json
• .env.example
• examples/tasks.sample.json

Typical Workflow

1. Add or import tasks into the local queue.
2. Let the scheduler process pending work on feature branches.
3. Review progress in the web dashboard or via Telegram.
4. When no tasks are pending, let scheduled QA analyze recent changes, architecture, docs, or performance.
5. Inspect generated logs and reports locally, then promote useful changes back to your main Unity repository.

More detail lives in:

• docs/setup.md
• docs/configuration.md
• docs/architecture.md
• docs/workflows.md
• docs/security.md

Screenshots

The repository does not currently ship static UI captures. Good additions for a public portfolio release would be:

• dashboard overview
• task detail page with diff preview
• QA report page
• Telegram command examples

Roadmap

• improve first-run bootstrap for local config creation
• add packaged dev dependencies for tests and linting
• add example screenshots and demo GIFs
• separate internal naming from public branding more cleanly
• expand QA prompt packs and public sample reports

Repository Layout

.
|-- daemon/          Python application code
|-- docs/            Public documentation
|-- examples/        Safe starter data
|-- skills/          Built-in review and Unity guidance rules
|-- run.py           Entrypoint
|-- requirements.txt Runtime dependencies

License

This repository is released under the MIT License. See LICENSE.