One developer + DevSquire = a full engineering team.
Automate your GitHub workflow — from issue to merged PR — without leaving VS Code.

English | 中文

The Problem

Developers spend ~40% of their time on repetitive workflow tasks: triaging issues, switching between browser and IDE, manually reviewing PRs, checking CI, and writing standup reports.

DevSquire eliminates that overhead. Point it at an issue, and it delivers a tested PR — autonomously.

Why DevSquire?

	Traditional AI Tools	DevSquire
Workflow	Copy-paste between chat and IDE	End-to-end autonomous pipeline
Issue → PR	Manual: read, plan, code, test, push	One click: analyze → implement → test → PR
Code Review	Generic suggestions	Structured review with severity levels
CI Monitoring	Manual checks and fixes	Auto-detect, auto-fix, auto-push
Visibility	Black box	Real-time dashboard with live progress

Core Features

🚀 Develop Issues

Issue in, PR out. Analyze → explore → plan → implement → test → create PR — fully autonomous.

🔍 Review PRs

Structured review with three modes: normal (interactive), auto (publish immediately), quick-approve (approve or block). Severity filtering: high / medium / low.

👁️ Watch PRs

Monitor open PRs continuously. Auto-detect CI failures and review comments, auto-fix and push — PRs stay green.

📊 Dashboard

Six tabs — Issues, PRs, Tasks, Actions, Skills, Report. Live pipeline progress, one-click actions.

📋 End-of-Day Report

One-click scrum summary: issues closed, PRs merged, open work.

Screenshots

Dashboard — Issue Management

Browse and develop GitHub issues directly from the dashboard. Task Pipeline — Real-Time Progress
Every task shows a live progress pipeline. Dev issues go through 7 stages; PR reviews through 3. Agent at Work — Terminal Integration
Tasks run in VS Code terminals. Click Terminal to see the AI agent working in real time. End-of-Day Report
Generate a daily scrum summary with one click.

Quick Start

1. Install

Search DevSquire in the VS Code Extensions Marketplace, or:

code --install-extension frankliu20.devsquire-vscode

2. Prerequisites

GitHub CLI (gh) — installed and authenticated
One AI backend:
- Claude Code — claude command
- GitHub Copilot CLI — gh copilot command

3. First Run

Open a GitHub-connected project in VS Code
Click the DevSquire icon in the sidebar to open the dashboard
Go to the Issues tab → pick an issue → click Develop
Watch the pipeline progress in the Tasks tab

Switch AI backends anytime via devSquire.aiPlatform setting.

How It Works

GitHub Issue ──► DevSquire Dashboard ──► AI Agent in Terminal
                        │                        │
                   real-time progress       autonomous pipeline
                   pipeline tracking        analyze → plan → code → test
                        │                        │
                        ▼                        ▼
                 Tasks Tab (live)          Pull Request Created

Task Type	Pipeline
Dev Issue	`Analyzing → Exploring → Planning → Implementing → Testing → Creating PR`
Review PR	`Reviewing → Done → Published`
Watch PR	`Monitoring ↔ Fixing CI ↔ Fixing Comments` (cyclic)

Settings

Setting	Default	Description
`devSquire.aiPlatform`	`claude-code`	AI backend: `claude-code` or `copilot-cli`
`devSquire.agentConfigLocation`	`home`	Agent config location: `home` or `project`
`devSquire.autoSyncAgentConfig`	`true`	Auto-sync agents on activation
`devSquire.devIssue.mode`	`auto`	`normal` (pause for approval) or `auto` (autonomous)
`devSquire.reviewPR.mode`	`normal`	`normal`, `auto`, or `quick-approve`
`devSquire.reviewPR.level`	`medium`	Review detail: `high`, `medium`, or `low`
`devSquire.watchPR.autoFixCI`	`true`	Auto-fix CI failures
`devSquire.watchPR.autoFixComments`	`false`	Auto-fix review comments

License

MIT — build something great with it.

DevSquire — Agentic Engineer