auto-bug-fix

Automates end-to-end mobile bug fixing from JIRA ticket through device reproduction to pull request.

Updated Apr 14, 2026
One-click install
npx skills add https://github.com/sunfmin/auto-bug-fix --skill auto-bug-fix-sunfmin
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: auto-bug-fix
Source: https://github.com/sunfmin/auto-bug-fix
Command: npx skills add https://github.com/sunfmin/auto-bug-fix --skill auto-bug-fix-sunfmin

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires curl, python3, ffmpeg.

What problem does it solve? Manually triaging mobile bugs requires reading JIRA tickets, reproducing issues on physical devices, hunting for root causes in source code, and writing fixes with evidence — a slow, repetitive cycle. This Skill automates the entire 8-phase workflow from ticket to verified pull request. ## Core Features & Use Cases - Autonomous Bug Reproduction: Parses JIRA tickets, executes reproduction steps on a real iOS/Android device via WebDriverAgent, and records video evidence using screenshot stitching with ffmpeg. - Root Cause Analysis and Fix: Traces the bug to specific files and lines, writes a failing integration test, applies a minimal fix, and verifies on-device with vision-based assertions. - Use Case: A QA engineer receives JIRA ticket PROJ-1234 reporting a wrong cart total. The Skill fetches the ticket, reproduces the bug on a connected iPhone with video evidence, finds the faulty calculation in the source, writes a failing test, fixes it, verifies on device, and opens a PR with the JIRA ticket updated. ## Quick Start Fix bug PROJ-1234 by reproducing it on my connected iPhone and creating a pull request with the fix.

Frequently Asked Questions about auto-bug-fix

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I automatically reproduce a mobile bug from a JIRA ticket?▼

Provide the JIRA ticket ID and the Skill fetches it via the Atlassian MCP, extracts reproduction steps, and executes them on a connected device using WebDriverAgent. It records video evidence by stitching screenshots with ffmpeg and uploads results back to JIRA.

How to automate iOS UI testing with WebDriverAgent?▼

Use the included wda.sh helper to create sessions, find elements by accessibility label, tap exact coordinates, and capture screenshots over the WebDriverAgent HTTP API on port 8100. Vision-based assertions evaluate screenshots instead of guessing coordinates.

Does this work with Android devices or only iOS?▼

The workflow targets both iOS and Android, but the provided wda.sh script and setup instructions are built around WebDriverAgent and xcodebuild for iOS. Android would require an equivalent UI automation layer such as UiAutomator2.

Why does the workflow write an integration test instead of a UI test?▼

Integration tests run in milliseconds, are deterministic without a device, and pinpoint the exact buggy code path. UI tests are reserved for the final verification phase where the fix is confirmed on a real device with vision.

When does the automated bug fix stop and ask for human review?▼

It stops when the ticket lacks concrete steps, the bug fails to reproduce after 3 attempts, the root cause cannot be tied to specific files and lines, or the fix exceeds 50 lines or 3 files. Partial findings are commented on JIRA for a human developer.