DevGenius – Intelligent, AI‑Powered Work Item Content & Task Generation for Azure DevOps

Why DevGenius?

DevGenius supercharges Agile teams by accelerating the breakdown of high‑level requirements into clear, executable work. Connected to a local Large Language Model (LLM – e.g., Ollama) or an external provider, it automatically generates structured task proposals (Task, Bug, User Story, Feature) and their descriptions directly from the context of the currently opened Work Item.

The result: faster backlog refinement, reduced planning friction, consistent quality of task descriptions across the organization – with full control over your data.

Key Benefits

• ⚡ Rapid Decomposition: Turn a User Story or Feature into a curated list of actionable tasks in seconds.
• 🤖 Local LLM Privacy: Point to your own on‑prem / local endpoint (e.g., Ollama) – keep sensitive project data inside your infrastructure.
• 🎯 Context‑Aware Generation: Uses the live Work Item title (and optionally your edits) for highly relevant output.
• 🛠 Multi‑Type Support: Save generated items as Task / Bug / User Story / Feature – you decide the granularity.
• 🌍 Multilingual Output: Choose the model response language (Polish, English, German, French, Spanish, Italian…).
• 🔐 Secure Configuration: Endpoints / tokens stored via Azure DevOps Extension Data – no hardcoding, no leaks.
• 🧩 Flexible Models: Use a predefined model or specify a custom one (e.g., gemma3:4b) with a simple setting.

How It Works (At a Glance)

1. Open a Work Item (e.g., User Story) in Azure DevOps.
2. Click the contextual "Create Tasks" button.
3. The dialog preloads the Work Item title – adjust it if needed.
4. Set the desired generation range (e.g., 4–8 items).
5. Pick the target Work Item type to create (Task / Bug / User Story / Feature).
6. Generate – the LLM returns a structured JSON list of tasks with descriptions & activity types.
7. Optionally edit any generated row (title, description, activity type).
8. Select the ones you want and click "Save selected" – they are created as children of the current Work Item.

Feature Highlights

Task Generation

• Adjustable min/max count (range input).
• Strict JSON contract ensures predictable parsing.
• Activity categorization (Development / Testing / Design / Documentation).

Editing & Selection

• Inline edit before committing tasks.
• Multi‑select and batch creation with one action.

Deep Azure DevOps Integration

• Parent/child linkage (System.Parent + Hierarchy relation).
• Dynamic Work Item type creation based on user selection.

Configuration Panel (DevGenius Settings)

• Provider choice: Local (Ollama / custom) or predefined models.
• Custom model name field (e.g., gemma3:4b, mistral, llama variants).
• Endpoint & token fields (only if the provider requires it).
• Enable/disable per Work Item type (visibility control).
• Model language selection separate from UI language.

Use Cases

• Sprint planning refinement right after writing a User Story.
• Breaking Epics into Features / Stories rapidly.
• Enforcing consistent descriptive standards across large teams.
• Helping onboard new team members with structured starter tasks.

Why a Local LLM?

• Your feature ideas and domain language never leave your network.
• Zero per‑request billing fees to external AI vendors.
• Easier domain adaptation via prompt engineering or local fine‑tuning.

Installation (Quick Start)

1. Install the extension from Azure DevOps Marketplace.
2. Open "DevGenius Settings" (Project Settings → Extensions → DevGenius).
3. Select provider (e.g., Ollama) and set: Endpoint URL, Model Name, Response Language.
4. Choose Work Item types where the button should appear.
5. Save settings.
6. Open a Work Item and start generating.

Requirements

• Azure DevOps Services or Azure DevOps Server (REST & Extension Data support).
• Appropriate permissions to manage extension settings.
• A reachable LLM endpoint (local Ollama REST API or other) if using a custom model.

Privacy & Security

• No data leaves your environment beyond the endpoint you explicitly configure.
• Tokens and endpoints are stored using Azure DevOps extension data services.
• You can operate fully locally – no dependency on third‑party SaaS AI.

FAQ

Can I use models other than Ollama presets?
Yes. Provide an HTTP endpoint that accepts your prompt and returns structured JSON.

Does it modify existing Work Items?
No. It creates new child items under the current parent.

Can I extend activity types?
Not yet – a customizable taxonomy is on the roadmap.

Does it work offline?
If your local LLM endpoint is accessible inside your network, yes – no external cloud required.

Roadmap

• Per‑Work‑Item‑Type prompt templates.
• Intelligent improvement suggestions for existing tasks.
• Automatic effort estimation powered by historical velocity.
• Tag & Area Path auto‑assignment rules.
• Extensible activity / classification dictionary.
• Bulk refinement mode for existing backlog items.

SEO / Marketplace Keywords

Azure DevOps, Work Item Automation, Task Generator, Backlog Refinement, User Story Decomposition, Agile Productivity, LLM, AI Assistant, Backlog Automation, Ollama, Local AI, Feature Planning, Sprint Planning, Dev Productivity

Quick Tips

• Start with a smaller range (3–5) for simple stories; increase for complex features.
• Tweak generated descriptions to add domain nuances before saving.
• Align on one language in the team for cleaner backlog search.

Support

Have a feature idea or found an issue? Open an issue via the extension support channel or (if available) the public repository. Real‑world feedback drives DevGenius forward.

DevGenius – accelerate planning, protect your data, and ship structured work items in seconds.