Conductor

Your AI Development Team: From spec to shipping code, automatically.

Conductor is a VS Code extension that orchestrates 7 AI agents to automate your entire development workflow—creating tasks from specifications, writing tests, reviewing quality, and fixing issues with one click.

What Conductor Does

Instead of just helping you code faster, Conductor automates the complete development workflow:

📋 Spec Validator - Your AI Project Manager

Reads your SPEC.md/REQUIREMENTS.md
Parses all MUST/SHOULD/MAY requirements
Creates SPEC-*.md issues for missing features
Ensures you never forget to implement a requirement
Turn specs into engineering backlogs automatically

🧪 Test Writer - Your AI QA Engineer

Analyzes your project structure and type
Writes only justified test types (context-aware, no over-engineering)
Small calculator → E2E only
E-commerce site → Full suite (E2E, unit, integration, security, performance, accessibility)
Creates TEST_STRATEGY.md explaining choices

🔄 Test Runner - Your Verification Agent

Executes all tests automatically
Creates ISSUE-*.md for each failure
Tracks test cycles and regressions
Generates ISSUE_SUMMARY.md overview

🎨 UX Reviewer - Your Design QA

Captures screenshots of all pages/components
Checks WCAG accessibility compliance
Reviews responsive design (mobile, tablet, desktop)
Creates UX-*.md issues with visual documentation
Generates UX_SUMMARY.md report

💻 Code Reviewer - Your Senior Engineer

Analyzes code patterns and best practices
Checks maintainability and technical debt
Creates CODE-*.md issues with suggestions
Generates CODE_REVIEW_SUMMARY.md

🛡️ Security Reviewer - Your Security Auditor

Context-aware security auditing (skips irrelevant checks)
OWASP Top 10 vulnerability scanning
Dependency vulnerability detection
Auth/session/PII validation
Creates SEC-*.md issues with severity ratings
Generates SECURITY_SUMMARY.md and SECURITY_POSTURE.md

🤖 Auto-Fixer - Your Fix Machine

Reads ALL issue types (test, UX, code, security, spec)
Fixes issues automatically with context-aware solutions
Updates issue status and adds fix notes
Validates fixes and handles regressions
The only QA tool that fixes what it finds

Why Conductor?

Complete Workflow Automation

Other tools help you code faster. Conductor manages your entire workflow.

GitHub Copilot writes code you ask for → Conductor creates tasks from specs AND fixes bugs
Linear/Jira tracks tasks you create → Conductor auto-creates AND auto-fixes them
Snyk/SonarQube detect issues → Conductor detects AND auto-fixes them

Context-Aware Intelligence

Not all projects need the same testing/security approach:

Calculator app → E2E tests + dependency checks (minimal)
E-commerce site → Full test suite + comprehensive security audit
Static blog → E2E + accessibility only

Conductor analyzes YOUR project and applies appropriate strategies.

One-Click Remediation

All issues appear in VS Code sidebar with:

"Fix This" button - Auto-fix individual issues
"Fix All" button - Auto-fix entire categories
Fix Notes - Detailed explanations of what changed

Quick Start

1. Install the Extension

From VS Code Marketplace:

Search for "Conductor" in Extensions
Click "Install"

From Source:

cd agentic-workflow-extension
npm install
npm run compile
# Press F5 to launch

2. Initialize in Your Project

On first activation, Conductor prompts:

Conductor is not set up in this project. Would you like to initialize it?
[ Yes, Initialize ] [ Not Now ]

Click "Yes, Initialize" to automatically create:

agentic-workflow/ folder with all 7 agent templates
Shell scripts for agent management
HANDOFF.md and CLAUDE.md templates
Quick Start guide

3. Write a Specification (Optional but Recommended)

Create SPEC.md in your project root:

# My SaaS App Specification

## Authentication
- MUST support email/password login
- MUST implement "Remember Me" functionality
- SHOULD support OAuth (Google, GitHub)
- MAY support 2FA

## Dashboard
- MUST display user's recent activity
- MUST be mobile responsive
- SHOULD load in <2 seconds

4. Start Building with Claude Code

claude

Tell Claude to read your Conductor commands:

You: Read CLAUDE.md to learn about Conductor commands

Now build your features:

You: Build a login page with email/password and a remember me checkbox

5. Run Quality Checks

Once you have code, run Conductor commands through Claude:

You: Conductor: spec-check
# Spec Validator creates SPEC-*.md issues for missing features (OAuth, 2FA, etc.)

You: Conductor: write tests
# Test Writer analyzes your login page and writes appropriate tests

You: Conductor: run tests
# Test Runner executes tests and creates ISSUE-*.md for failures

You: Conductor: ux-review
# UX Reviewer captures screenshots and creates UX-*.md for accessibility issues

You: Conductor: code-review
# Code Reviewer analyzes code quality and creates CODE-*.md issues

You: Conductor: security-review
# Security Reviewer audits for vulnerabilities and creates SEC-*.md issues

You: Conductor: fix all
# Auto-Fixer reads all open issues and fixes them automatically

6. View and Fix Issues in VS Code

Open the "Conductor" panel in VS Code sidebar
See issues grouped by type:
- 🛡️ Security Issues (SEC-*.md)
- 🧪 Test Issues (ISSUE-*.md)
- 📋 Spec Issues (SPEC-*.md)
- 🎨 UX Issues (UX-*.md)
- 💻 Code Quality (CODE-*.md)
Click "Fix This" to auto-fix individual issues
Click "Fix All" to fix entire categories
View "Fix Notes" to see what changed

Commands Reference

All Conductor commands work through Claude Code:

Command	What It Does
`Conductor: spec-check`	Validate entire specification
`Conductor: spec-check:scope [what]`	Validate specific spec section
`Conductor: write tests`	Write context-aware tests
`Conductor: write tests:scope [what]`	Write tests for specific features
`Conductor: run tests`	Execute tests and create issues
`Conductor: ux-review`	Review UX/accessibility
`Conductor: ux-review:scope [what]`	Review specific pages/flows
`Conductor: code-review`	Analyze code quality
`Conductor: code-review:scope [what]`	Review specific modules
`Conductor: security-review`	Audit security vulnerabilities
`Conductor: security-review:scope [what]`	Audit specific modules
`Conductor: fix all`	Auto-fix all open issues
`Conductor: fix:scope [what]`	Fix issues matching scope

Use :scope for large codebases to focus agents on specific features/files.

Real-World Examples

Solo Developer

"I wrote a 2-page spec for my SaaS. Conductor created 47 tasks, wrote 120 tests, found 12 security issues, and fixed them all. Saved me 2 weeks."

Before Conductor:

Manually tracked requirements → forgot 3 critical features
Spent 2 weeks writing tests
Shipped with accessibility issues
Had 12 security vulnerabilities in production

After Conductor:

Spec Validator created 47 tasks from spec automatically
Test Writer wrote 120 tests in 10 minutes
UX Reviewer found 33 accessibility issues before launch
Security Reviewer caught all 12 vulnerabilities
Auto-Fixer resolved everything with one click

Agency

"We use Conductor on client projects. Spec Validator ensures we don't miss requirements. Zero client complaints."

Before Conductor:

Client complained we missed 8 requirements → $5k refund
Manual QA missed mobile responsiveness bug
Spent 40 hours/project on testing

After Conductor:

100% requirement coverage (Spec Validator)
UX Reviewer catches responsive issues automatically
Testing time reduced from 40 → 5 hours
Zero client complaints

Open Source Maintainer

"I run Conductor on every PR. Contributors love the auto-fix feature. Code review time: 10 hours → 1 hour/week."

Before Conductor:

60% of PRs had no tests
Manual review of every contribution
Accessibility issues slipped through
10 hours/week on code review

After Conductor:

Test Writer generates tests for contributor code
UX/Security reviews happen automatically
Contributors auto-fix their own issues
1 hour/week on code review

Issue Lifecycle

Issues flow through these statuses:

open → fixed → verified
  ↓
regression → open

Status Indicators:

🔴 Open - Needs to be fixed
✅ Fixed - Has been resolved
✅ Verified - Fix has been tested and confirmed
🔄 Regression - Previously fixed, but broken again

Severity Levels:

🔴 Critical - Must fix immediately
🟠 Major/High - Should fix soon
🟡 Minor/Medium - Nice to have
💡 Suggestion - Optional improvement

Visual Issue Management

All issues appear in the Conductor sidebar:

Conductor
├── 🛡️ Security Issues (3 open)
│   ├── 🔴 SEC-001: SQL injection in login form
│   ├── 🟠 SEC-002: Missing CSRF tokens
│   └── 🟡 SEC-003: Weak password requirements
├── 🧪 Test Issues (5 open)
│   ├── 🔴 ISSUE-001: Login form validation failing
│   └── ...
├── 📋 Spec Issues (2 open)
│   ├── 🔴 SPEC-001: Missing OAuth integration
│   └── 🟡 SPEC-002: 2FA not implemented (MAY requirement)
├── 🎨 UX Issues (8 open)
│   ├── 🔴 UX-001: Missing ARIA labels
│   └── ...
└── 💻 Code Quality (4 open)
    ├── 🟠 CODE-001: Duplicate validation logic
    └── ...

Right-click any issue:

Fix This Issue
Fix All Issues (in group)
Reject Issue
Revert Fix
Show Fix Notes
Open Issue File

Context-Aware Intelligence

Smart Test Strategy

Conductor analyzes your project and writes only justified tests:

Calculator App (10 files):

✅ E2E tests (verify calculations work)
❌ No unit tests (over-engineering)
❌ No performance tests (not needed)

E-Commerce Site (500+ files):

✅ E2E tests (user flows)
✅ Unit tests (business logic)
✅ Integration tests (payment, inventory)
✅ Security tests (auth, PII)
✅ Performance tests (load handling)
✅ Accessibility tests (WCAG compliance)

REST API (no UI):

✅ Unit tests
✅ Integration tests
✅ API contract tests
✅ Security tests
❌ No E2E UI tests (no UI)

Creates TEST_STRATEGY.md explaining which test types are justified for YOUR project.

Smart Security Auditing

Conductor analyzes your project's risk profile:

Calculator App:

✅ Dependency vulnerability checks
❌ Skip auth/session checks (no auth)
❌ Skip SQL injection checks (no database)
❌ Skip PII checks (no user data)

E-Commerce Site:

✅ Full OWASP Top 10 audit
✅ Auth/session security
✅ Payment security (PCI-DSS considerations)
✅ PII handling (GDPR compliance)
✅ SQL injection, XSS, CSRF

Static Blog:

✅ Dependency checks
❌ Skip backend vulnerabilities (no backend)

Creates SECURITY_POSTURE.md explaining which audits are relevant.

Scoped Commands (For Large Codebases)

Focus agents on specific features instead of reviewing the entire codebase:

# Write tests only for authentication
You: Conductor: write tests:scope authentication

# Review UX only for checkout flow
You: Conductor: ux-review:scope checkout flow

# Review only authentication code
You: Conductor: code-review:scope authentication

# Fix only payment-related issues
You: Conductor: fix:scope payment

When to use scoping:

Large codebases (100+ files)
Working on specific feature
Incrementally reviewing new code
Debugging particular user flow

File Structure

Conductor creates this structure automatically:

your-project/
├── SPEC.md                              # Your requirements (optional)
├── HANDOFF.md                           # Project context for agents
├── CLAUDE.md                            # Conductor command reference
├── TEST_STRATEGY.md                     # Which tests are justified (generated)
├── TEST_PLAN.md                         # Test coverage details (generated)
├── SECURITY_POSTURE.md                  # Security audit scope (generated)
├── ISSUE_SUMMARY.md                     # Test failures overview (generated)
├── UX_SUMMARY.md                        # UX issues overview (generated)
├── CODE_REVIEW_SUMMARY.md               # Code quality overview (generated)
├── SECURITY_SUMMARY.md                  # Security audit report (generated)
├── SPEC_COVERAGE.md                     # Spec compliance report (generated)
└── agentic-workflow/
    ├── .agents/                         # Agent prompt templates
    │   ├── spec-validator/
    │   ├── test-writer/
    │   ├── test-runner/
    │   ├── ux-reviewer/
    │   ├── code-reviewer/
    │   ├── security-reviewer/
    │   └── auto-fixer/
    ├── .issues/                         # Generated issues
    │   ├── spec/SPEC-*.md
    │   ├── test/ISSUE-*.md
    │   ├── ux/UX-*.md
    │   ├── code/CODE-*.md
    │   └── security/SEC-*.md
    ├── .test_state/                     # Agent state tracking
    ├── start-agent.sh                   # Launch individual agents
    ├── start-all.sh                     # Launch all agents in tmux
    └── mega-orchestrator.sh             # Deterministic workflow

Integration with Claude Code

Conductor is designed to work seamlessly with Claude Code CLI:

Write code with Claude
Run Conductor commands through Claude
Fix issues displayed in VS Code sidebar
Repeat until all issues resolved

Example workflow:

claude

You: Read CLAUDE.md
You: Build a login page
You: Conductor: write tests
You: Conductor: run tests
You: Conductor: fix all

VS Code Commands

All commands available via Cmd+Shift+P:

Conductor: Refresh Issues - Reload issue tree
Conductor: Show Issues Panel - Focus on sidebar
Conductor: Show Open Issues Only - Filter to open
Conductor: Show Fixed Issues Only - Filter to resolved
Conductor: Update Agent Templates - Manually update templates

Template Updates

Conductor automatically checks for template updates on activation:

Detects version mismatch: "New agent templates available (v1.1.0). Update now?"
Backs up your customized agents to .agents.backup/
Copies new templates and shell scripts
Updates version file

Manual update: Cmd+Shift+P → "Conductor: Update Agent Templates"

Troubleshooting

Panel is empty

Ensure you have agentic-workflow/.issues/ folder
Run Conductor commands through Claude to create issues
Check Output panel: View → Output → "Conductor"

Issues not updating

Click the refresh button (↻) in the panel
Check file permissions on .issues/ folder
Restart VS Code

Can't see the panel

View → Open View → "Conductor"
Or search "Conductor" in the Explorer sidebar

Extension not activating

Check workspace has agentic-workflow/ folder
View → Output → "Conductor" for errors
Cmd+Shift+P → "Developer: Reload Window"

Requirements

VS Code 1.85.0 or higher
Claude Code CLI (recommended for best experience)
Node.js (for running agents)

Development

Project Structure

agentic-workflow-extension/
├── src/
│   ├── extension.ts        # Main activation & commands
│   ├── issueProvider.ts    # TreeView data provider
│   └── issueParser.ts      # Markdown frontmatter parser
├── templates/              # Bundled agent templates
│   ├── .agents/
│   ├── start-agent.sh
│   ├── start-all.sh
│   └── mega-orchestrator.sh
├── package.json            # Extension manifest
├── tsconfig.json           # TypeScript config
└── README.md               # This file

Build Commands

npm run compile     # Compile TypeScript
npm run watch       # Watch mode for development
npm run package     # Create .vsix for distribution

Testing

Press F5 in VS Code to launch Extension Development Host window.

Roadmap

Phase 1 (Current - Free)

[x] 7 AI agents (spec, test, UX, code, security, auto-fix)
[x] Visual issue management
[x] One-click auto-fix
[x] Context-aware intelligence
[x] Automatic setup and updates
[x] Real-time file watching

Phase 2 (Planned)

[ ] Team features (shared configs, sync)
[ ] Advanced filtering and search
[ ] Export issues (CSV, JSON, GitHub Issues)
[ ] Statistics dashboard
[ ] CI/CD integrations (GitHub Actions, GitLab)
[ ] Slack/Discord notifications

Phase 3 (Future)

[ ] Custom agent builder
[ ] Cloud sync
[ ] Team analytics
[ ] Enterprise features (SSO, audit logs)

License

MIT

Contributing

Issues and PRs welcome!

Report bugs: GitHub Issues
Feature requests: GitHub Discussions
Join community: Discord

Learn More

Documentation: Full docs
Examples: Sample projects
Blog: Conductor blog
Twitter: @conductor_dev

Credits

Built with ❤️ by developers who hate manual QA.

Ready to ship quality code faster? Install Conductor and automate your entire development workflow.

Install from VS Code Marketplace

Conductor

Mosaik Digital

Conductor

What Conductor Does

📋 Spec Validator - Your AI Project Manager

🧪 Test Writer - Your AI QA Engineer

🔄 Test Runner - Your Verification Agent

🎨 UX Reviewer - Your Design QA

💻 Code Reviewer - Your Senior Engineer

🛡️ Security Reviewer - Your Security Auditor

🤖 Auto-Fixer - Your Fix Machine

Why Conductor?

Complete Workflow Automation

Context-Aware Intelligence

One-Click Remediation

Quick Start

1. Install the Extension

2. Initialize in Your Project

3. Write a Specification (Optional but Recommended)

4. Start Building with Claude Code

5. Run Quality Checks

6. View and Fix Issues in VS Code

Commands Reference

Real-World Examples

Solo Developer

Agency

Open Source Maintainer

Issue Lifecycle

Visual Issue Management

Context-Aware Intelligence

Smart Test Strategy

Smart Security Auditing

Scoped Commands (For Large Codebases)

File Structure

Integration with Claude Code

VS Code Commands

Template Updates

Troubleshooting

Panel is empty

Issues not updating

Can't see the panel

Extension not activating

Requirements

Development

Project Structure

Build Commands

Testing

Roadmap

Phase 1 (Current - Free)

Phase 2 (Planned)

Phase 3 (Future)

License

Contributing

Learn More

Credits