/init

/init - Initialize AI Dev Team for a Repository

When this command is invoked, perform the following steps IN ORDER. Use the Bash and Write tools to actually create files and directories.

Step 0: Check for Existing Configuration

First, check if .ai-dev-team/config/project.json already exists.

If it exists, ask the user:

Existing AI Dev Team configuration found.

What would you like to do?
1. Reconfigure everything (overwrites existing files)
2. Keep existing config, just create missing directories
3. Cancel

> [User selects]

• If user selects 1: Continue with all steps, overwriting files
• If user selects 2: Skip to directory creation, don't overwrite config files
• If user selects 3: Stop and do nothing

Step 1: Create Directory Structure

Execute these commands:

mkdir -p .ai-dev-team/config
mkdir -p .ai-dev-team/workflow/active
mkdir -p .ai-dev-team/workflow/archive
mkdir -p .ai-dev-team/context/project

Step 2: Create .ai-dev-team/.gitignore

Write this file:

Path: .ai-dev-team/.gitignore

.env
*.local.json

Step 3: Create .ai-dev-team/.env

Write this file:

Path: .ai-dev-team/.env

# AI Dev Team - User Configuration
# Run Atlassian:atlassianUserInfo to get your account ID

ATLASSIAN_ACCOUNT_ID=
STAKEHOLDER_NAME=
TEAM_ROLE=stakeholder

Then ask the user for their name and use Atlassian:atlassianUserInfo to get their account ID. Update the .env file with these values.

Step 4: Create .ai-dev-team/config/project.json

Write this file:

Path: .ai-dev-team/config/project.json

{
  "plugin": {
    "source": "marketplace",
    "name": "ai-dev-team"
  },
  "project": {
    "name": "",
    "code": "",
    "description": ""
  },
  "jira": {
    "cloudId": "",
    "projectKey": "",
    "projectId": "",
    "issueTypes": {
      "epic": null,
      "story": null,
      "task": null,
      "bug": null,
      "subtask": null
    }
  },
  "confluence": {
    "cloudId": "",
    "spaceId": "",
    "spaceKey": ""
  },
  "github": {
    "owner": "",
    "repo": "",
    "defaultBranch": "main"
  }
}

Then interactively fill in the values:

1. Ask for project name and code
2. Run Atlassian:getAccessibleAtlassianResources to get Cloud IDs
3. Run Atlassian:getVisibleJiraProjects to list projects, let user select
4. Run Atlassian:getJiraProjectIssueTypesMetadata to get issue type IDs
5. Run Atlassian:getConfluenceSpaces to list spaces, let user select
6. Ask about GitHub repo (optional)

Update project.json with all gathered values.

Step 5: Create .ai-dev-team/CLAUDE.md

Write this file:

Path: .ai-dev-team/CLAUDE.md

# AI Dev Team Plugin Instructions

This project uses the AI Dev Team plugin for Agile workflow automation.

## Agent Activation

When the user requests work, activate the appropriate agent:

| User Request Pattern | Primary Agent | Description |
|---------------------|---------------|-------------|
| "I need a feature for..." | Orchestrator | Routes to Business Analyst |
| "Create stories for..." | Business Analyst | Creates user stories |
| "Design the architecture..." | Software Architect | Technical specs |
| "Implement..." | Senior Developer | Code implementation |
| "Review this code..." | Senior Developer | Code review |
| "Add tests to..." | Senior Developer / QA Engineer | Test creation |
| "Test..." | QA Engineer | Test execution |
| "Deploy..." | DevOps Engineer | Deployment |
| "What's the sprint status?" | Project Manager | Status reporting |

## How to Load Agents

When an agent is needed, read its definition from the ai-dev-team plugin.
The plugin is installed at: `~/.claude/plugins/marketplaces/ai-dev-team-marketplace/`

- Agents: `{plugin-path}/agents/{agent-name}.md`
- Skills: `{plugin-path}/skills/{category}/{skill-name}/SKILL.md`
- Workflows: `{plugin-path}/workflows/{workflow-name}.md`

## Project Configuration

- Config: `.ai-dev-team/config/project.json`
- User settings: `.ai-dev-team/.env`
- Active workflows: `.ai-dev-team/workflow/active/`
- Context: `.ai-dev-team/context/`

## Workflow Tracking

When running a workflow:
1. Create a folder in `.ai-dev-team/workflow/active/{workflow-type}-{identifier}/`
2. Create `workflow.md` with current status
3. Create `state.json` for tracking progress
4. Move to `.ai-dev-team/workflow/archive/` when complete

Step 5b: Update Root CLAUDE.md

Prepend this reference to the TOP of the existing CLAUDE.md:

If CLAUDE.md exists at the repo root, prepend this line at the very top:

<!-- AI Dev Team: If .ai-dev-team/CLAUDE.md exists, read and follow those instructions. -->

If CLAUDE.md does NOT exist, create it with just that line:

<!-- AI Dev Team: If .ai-dev-team/CLAUDE.md exists, read and follow those instructions. -->

Step 6: Create .ai-dev-team/context/README.md

Write this file:

Path: .ai-dev-team/context/README.md

# Context Directory

This directory stores project context summaries for AI agents.

## Structure

- `project/` - Project-level context
  - `solution-summary.md` - High-level solution overview
  - `skills-summary.md` - Available skills reference
  - `workflows-summary.md` - Workflow documentation

## Usage

Agents read from this directory to understand project context.
Context files are updated during workflows and after significant changes.

Step 7: Update Root .gitignore

Append to .gitignore (create if doesn't exist):

# AI Dev Team
.ai-dev-team/.env

Step 8: Confirmation

After all files are created, show the user:

## AI Dev Team Initialized!

Created:
- .ai-dev-team/config/project.json
- .ai-dev-team/.env
- .ai-dev-team/.gitignore
- .ai-dev-team/CLAUDE.md
- .ai-dev-team/workflow/active/
- .ai-dev-team/workflow/archive/
- .ai-dev-team/context/project/
- Updated CLAUDE.md with plugin reference

Configuration:
- Project: {project.name} ({project.code})
- Jira: {jira.projectKey}
- Confluence: {confluence.spaceKey}

Next steps:
1. Run `/story` to create your first user story
2. Say "I need a feature for..." to engage the team
3. Run `/workflow-dev` to start a development workflow

Error Handling

• If Atlassian MCP is not available, ask user to configure it first
• If directories already exist, ask if user wants to reconfigure
• If any step fails, show error and continue with remaining steps