Agent skill

Conductor Methodology

This skill should be used when the user asks about "conductor methodology", "context-driven development", "conductor directory structure", "conductor files", "tracks.md", "plan.md", "spec.md", or when commands reference conductor concepts. Provides comprehensive understanding of the Conductor framework for Context-Driven Development.

Stars 163
Forks 31

Install this agent skill to your Project

npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/data/conductor-methodology

SKILL.md

Conductor Methodology

Conductor implements Context-Driven Development, a methodology that treats project context as a first-class artifact alongside code. This skill provides essential knowledge about the Conductor framework, directory structure, file formats, and core concepts.

Core Philosophy

Measure twice, code once. Conductor enforces a strict workflow:

  1. Context: Establish project-level context once (product vision, tech stack, workflow)
  2. Spec & Plan: Create detailed specifications and hierarchical plans for each unit of work
  3. Implement: Follow the plan, update status, verify at phase boundaries

This approach ensures AI agents follow consistent, high-quality development practices with proper planning and context awareness.

Directory Structure

When Conductor is active in a project, it creates and manages this structure:

conductor/
├── product.md              # Product vision, users, goals
├── product-guidelines.md   # Brand voice, design principles, UX guidelines
├── tech-stack.md          # Languages, frameworks, architecture, tools
├── workflow.md            # Development methodology (TDD, commits, testing)
├── code_styleguides/      # Language-specific style guides
│   ├── general.md
│   ├── javascript.md
│   ├── typescript.md
│   ├── python.md
│   ├── go.md
│   └── html-css.md
├── tracks.md              # Master list of all tracks
└── tracks/
    └── <track_id>/
        ├── metadata.json  # Track metadata (ID, title, type, status, dates)
        ├── spec.md        # Detailed requirements and acceptance criteria
        └── plan.md        # Hierarchical implementation plan

Key Concepts

Tracks

A track is a logical unit of work - a feature, bug fix, or chore. Each track has:

  • Unique ID (timestamp-based: track_20231215_143022)
  • Type: feature, bug, or chore
  • Status: pending, in-progress, completed, or blocked
  • Dedicated directory with spec and plan

Specifications (spec.md)

Detailed requirements document for a track, including:

  • Overview and context
  • Functional requirements
  • Non-functional requirements (performance, security, etc.)
  • Acceptance criteria (definition of done)
  • Out of scope (what this does NOT include)

Plans (plan.md)

Hierarchical implementation plan with three levels:

  1. Phases: Major stages of work (e.g., "Backend API", "Frontend UI")
  2. Tasks: Specific actions within phases (e.g., "Create user model")
  3. Sub-tasks: Detailed steps within tasks (e.g., "Add email field")

Each item has a status marker:

  • [ ] - Pending (not started)
  • [~] - In Progress (currently working)
  • [x] - Completed (done)
  • [!] - Blocked (cannot proceed)

Workflow Methodology

The workflow.md file defines the development process, including:

  • Development methodology: TDD, BDD, or standard
  • Commit strategy: Conventional commits, descriptive messages
  • Testing requirements: Unit, integration, E2E
  • Code review: Required, optional, pair programming
  • Phase completion protocol: Verification steps at phase boundaries

File Formats

tracks.md Format

Master list of all tracks:

markdown
# Conductor Tracks

## Active Tracks

### [ ] Track: Add user authentication
- **ID**: track_20231215_143022
- **Type**: Feature
- **Status**: In Progress
- **Created**: 2023-12-15

### [ ] Track: Fix login redirect bug
- **ID**: track_20231215_150033
- **Type**: Bug
- **Status**: Pending
- **Created**: 2023-12-15

## Completed Tracks

### [x] Track: Setup project infrastructure
- **ID**: track_20231214_120000
- **Type**: Chore
- **Status**: Completed
- **Created**: 2023-12-14
- **Completed**: 2023-12-15

metadata.json Format

Track metadata in JSON:

json
{
  "track_id": "track_20231215_143022",
  "title": "Add user authentication",
  "type": "feature",
  "status": "in-progress",
  "created_at": "2023-12-15T14:30:22Z",
  "updated_at": "2023-12-15T16:45:00Z"
}

plan.md Format

Hierarchical plan with status markers:

markdown
# Implementation Plan: Add User Authentication

## Phase 1: Backend API

### [ ] Task: Create user model
- [ ] Sub-task: Define user schema
- [ ] Sub-task: Add email and password fields
- [ ] Sub-task: Add timestamps

### [~] Task: Implement authentication endpoints
- [x] Sub-task: Create /register endpoint
- [~] Sub-task: Create /login endpoint
- [ ] Sub-task: Create /logout endpoint

### [ ] Task: Conductor - User Manual Verification 'Backend API' (Protocol in workflow.md)

## Phase 2: Frontend UI

### [ ] Task: Create login form
- [ ] Sub-task: Design form layout
- [ ] Sub-task: Add form validation
- [ ] Sub-task: Connect to API

### [ ] Task: Conductor - User Manual Verification 'Frontend UI' (Protocol in workflow.md)

Project Maturity Types

Brownfield Projects

Existing projects with code, dependencies, or version control. Conductor analyzes:

  • Programming languages and frameworks
  • Architecture patterns
  • Project goals from README or manifests
  • Existing conventions and patterns

Pre-populates context files based on analysis.

Greenfield Projects

New projects starting from scratch. Conductor guides through:

  • Defining product vision
  • Selecting tech stack
  • Establishing workflow
  • Creating initial structure

Context Files

product.md

Defines product vision and goals:

  • Target users and their needs
  • Core features and capabilities
  • Product goals and success metrics
  • Constraints and assumptions

product-guidelines.md

Establishes brand and design standards:

  • Brand voice and tone
  • Design principles
  • UX guidelines
  • Content guidelines

tech-stack.md

Documents technical decisions:

  • Programming languages
  • Frontend and backend frameworks
  • Database and storage
  • Architecture patterns
  • Development tools

workflow.md

Defines development process:

  • Methodology (TDD, BDD, standard)
  • Commit conventions
  • Testing strategy
  • Code review process
  • Phase completion verification

Status Markers

Use consistent markers throughout plans:

Marker Status Meaning
[ ] Pending Not started
[~] In Progress Currently working
[x] Completed Done
[!] Blocked Cannot proceed

Phase Completion Verification

If workflow.md defines a "Phase Completion Verification and Checkpointing Protocol", each phase must include a final verification task:

markdown
### [ ] Task: Conductor - User Manual Verification '<Phase Name>' (Protocol in workflow.md)

This ensures quality gates at phase boundaries.

Commands Overview

Conductor provides five commands:

  1. setup: Initialize conductor environment (run once per project)
  2. newTrack: Create new track with spec and plan
  3. implement: Execute tasks from plan
  4. status: Display progress overview
  5. revert: Git-aware revert of tracks/phases/tasks

Best Practices

When Creating Specs

  • Be specific about requirements
  • Include acceptance criteria
  • Define what's out of scope
  • Reference product.md for context
  • Consider non-functional requirements

When Creating Plans

  • Break work into phases (major stages)
  • Define tasks within phases (specific actions)
  • Add sub-tasks for detailed steps
  • Follow workflow methodology (e.g., TDD tasks)
  • Include phase completion verification tasks
  • Keep tasks atomic and testable

When Implementing

  • Load all context files before starting
  • Follow the plan sequentially
  • Update status markers in real-time
  • Commit frequently (per task or phase)
  • Run automated checks at phase boundaries
  • Verify completion before moving to next phase

When Updating Context

  • Keep context files synchronized with reality
  • Update tech-stack.md when adding new technologies
  • Update workflow.md when process changes
  • Update product.md when goals evolve
  • Commit context changes separately

Integration with Git

Conductor is Git-aware:

  • Tracks commits associated with tasks
  • Records commit SHAs in plan.md
  • Enables intelligent revert of logical units
  • Handles non-linear history (rebases, squashes)

Additional Resources

For detailed patterns and advanced techniques, consult:

  • references/brownfield-analysis.md - Brownfield project analysis patterns
  • references/plan-generation.md - Plan generation strategies
  • examples/sample-track/ - Complete example track

Common Patterns

Starting a New Project

  1. Run /claude-conductor:setup
  2. Answer questions about product, tech stack, workflow
  3. Review generated context files
  4. Create first track with /claude-conductor:newTrack

Working on a Feature

  1. Create track: /claude-conductor:newTrack "Feature description"
  2. Review and approve spec
  3. Review and approve plan
  4. Implement: /claude-conductor:implement
  5. Verify at phase boundaries
  6. Complete track

Checking Progress

  1. Run /claude-conductor:status
  2. Review active tracks
  3. Check completion percentages
  4. Identify blockers

Reverting Work

  1. Run /claude-conductor:revert
  2. Select track/phase/task to revert
  3. Review Git commits to be reverted
  4. Confirm and execute

Tips for Commands

When commands reference conductor concepts, use this skill to understand:

  • File formats and structure
  • Status marker meanings
  • Phase completion requirements
  • Context file purposes
  • Track lifecycle

This knowledge enables accurate implementation of conductor workflows and proper maintenance of conductor artifacts.

Didn't find tool you were looking for?

Be as detailed as possible for better results