Skip to content

Github.mdx

Latest commit

 

History

History
196 lines (157 loc) · 6.2 KB

Github.mdx

File metadata and controls

196 lines (157 loc) · 6.2 KB
Raw
title GitHub
icon github
description Create GitHub issues using voice commands through your OMI device
Install the GitHub Issues app from the Omi App Store Check out the full source code on GitHub

✨ Features

  • 🎤 Voice-Activated - Say "Feedback Post" and describe your problem
  • 🧠 AI-Powered - Collects 5 segments for detailed feedback, generates title + description
  • 🔐 OAuth Authentication - Secure GitHub OAuth 2.0 integration
  • 📦 Repository Selection - Choose which repo receives issues during setup
  • ⚙️ Flexible Settings - Change target repository anytime from mobile-first homepage
  • 🤖 Smart Formatting - AI extracts key info and formats professional issues
  • 🔕 Silent Collection - Only notifies when issue is created
  • 📱 Mobile-First UI - Beautiful responsive design for all devices

🚀 Quick Start

Install the [GitHub Issues app](https://h.omi.me/apps/01K7BHKJR79KQQW7ZZGZX7GCGD) in your OMI mobile app Authenticate your GitHub account (one-time setup) Choose the repository where issues should be created - Say: "Feedback Post, the app keeps crashing when I upload photos..." - Keep describing the problem naturally for 15-20 seconds - AI processes your speech and creates a formatted issue - You get a notification with the issue link! ✅

Trigger Phrases

- "Feedback Post" - "Create Issue" - "Report Issue" - "File Issue" - "New Issue"

How It Works

The app intelligently collects and processes your feedback:

  1. Detects "Feedback Post" → Starts collecting
  2. Automatically collects 5 segments (15-20 seconds of detailed speech)
  3. Sends all segments to AI
  4. AI generates professional title and detailed description
  5. Creates GitHub issue with "voice-feedback" label
  6. Notifies you with issue link! 🎉
**Example Workflow:**

You: "Feedback Post, the search function isn't working"
[collecting silently...]

You: "when I type in the search bar nothing happens"
[collecting silently...]

You: "I've tried on both Chrome and Safari"
[collecting silently...]

You: "and it worked fine last week but now"
[collecting silently...]

You: "it's completely broken on all browsers"
→ AI processes all 5 segments

AI Generated Issue:

  • Title: "Search function not working across browsers"
  • Description: The search function is currently not working. When typing in the search bar, no results appear. This issue has been tested on both Chrome and Safari browsers. The search functionality was working correctly last week but is now completely non-functional across all browsers.

→ Issue created! 🔔

🧠 AI Processing

The app uses OpenAI for intelligent issue generation:

  1. Collection - Gathers 5 segments (~15-20 seconds) for detailed context
  2. Title Generation - Creates concise, descriptive title
  3. Description Formatting - Structures problem statement professionally
  4. Cleanup - Removes filler words, fixes grammar, adds proper formatting

Example Transformation:

```text Voice Input (5 segments) "the app keeps crashing when I um try to upload photos it happens every single time on my iPhone 14 like the app just freezes for a second and then it completely closes and this started happening after the latest update yesterday" ``` 
Title: App crashes when uploading photos on iPhone 14

Description:
The app consistently crashes during photo uploads. When attempting to 
upload a photo, the app freezes briefly and then closes completely. 
This issue occurs every time on iPhone 14 and started after the latest 
update.

Labels: voice-feedback

📊 How Segments Work

OMI sends transcripts in segments as you speak. The app:

  • ✅ Detects "Feedback Post" trigger
  • ✅ Collects exactly 5 segments
  • ✅ Silent during collection (no spam)
  • ✅ AI processes all 5 together
  • ✅ One notification on completion

Why 5 segments?

  • Allows detailed problem description
  • Captures ~15-20 seconds of speech
  • Gives context for better issue formatting
  • AI has full information for title/description generation

Refreshing Repository List

Click "Refresh Repos" to:

  • Fetch latest list from GitHub
  • Include newly created repositories
  • Update repository access permissions

🔐 Security & Privacy

**Security Features:** - OAuth 2.0 authentication (no password storage) - Tokens stored securely with file persistence - Per-user token isolation - HTTPS enforced in production - State parameter for CSRF protection - Secure scope: only `repo` access

🐛 Troubleshooting

- Complete the GitHub OAuth flow - Try re-authenticating if the issue persists - Visit the app homepage - Select a repository from the dropdown menu - Click "Save Repository" to confirm - Verify the repository exists and you have write access - Ensure OAuth app has correct permissions - Click "Refresh Repos" button on the homepage - Ensure you have push access to the repository - Verify GitHub OAuth app permissions are correctly set

Support & Resources

Integration App Guide View the full codebase