Skip to content

Getting Started

Prerequisites

  • Agent Skills–compatible environment (e.g. Cursor IDE)
  • Jira or Azure DevOps – Active account with API access
  • GitHub – Account with personal access token

Step 1: Configure MCP

MCP combinations: - GitHub + Jira — GitHub for repos/PRs, Jira for issue tracking - ADO only — Azure DevOps provides both repo management and issue tracking

Full MCP setup → (Option A or B).

Open your environment’s MCP settings (e.g. Cursor: Settings → Features → Model Context Protocol) and add:

{
  "mcpServers": {
    "github": {
      "url": "https://api.githubcopilot.com/mcp/",
      "headers": {
        "Authorization": "Bearer <TOKEN>"
      }
    },
    "atlassian": {
      "url": "https://mcp.atlassian.com/v1/sse"
    },
  }
}

Get GitHub Token:

  1. Go to github.com/settings/tokens
  2. Generate new token (classic)
  3. Select scopes: repo and read:org
  4. Copy token into config above

Authorize Atlassian:

First time you use Jira, your environment may prompt for OAuth (e.g. Cursor prompts for Atlassian authorization).

Azure DevOps Alternative

For Azure DevOps instead of Jira, use this configuration:

"ado": {
  "command": "npx",
  "args": [
    "-y",
    "--registry",
    "https://registry.npmjs.org/",
    "@azure-devops/mcp",
    "your-organization-name"
  ]
}

Replace your-organization-name with your ADO organization.

Detailed MCP setup →

Step 2: Install Skills

This repo uses the Agent Skills format. The same skills/ layout and SKILL.md files work for Cursor, Claude, and Codex—each environment loads from its own skills directory (see table below). Cursor users can also install directly from GitHub; others copy from a release.

Environment Project-level User-level (global)
Cursor .cursor/skills/ ~/.cursor/skills/
Claude .claude/skills/ ~/.claude/skills/
Codex .codex/skills/ ~/.codex/skills/

Choose one of the following.

Option A: Install from GitHub (Cursor)

Cursor supports loading skills from a GitHub repository:

  1. Open Cursor Settings (e.g. Cmd+Shift+J / Ctrl+Shift+J) → Rules
  2. Under Project Rules, click Add Rule
  3. Select Remote Rule (Github)
  4. Enter this repository URL: 
    https://github.com/fancybread-com/sdlc-workflow-skills

Skills are loaded from the repo; you get updates when Cursor refreshes. No need to download or copy files.

Layout (Cursor, Claude, Codex)

The same layout works for all three: a top-level skills/ directory; each skill is a subfolder containing SKILL.md with YAML frontmatter name (matching the folder name) and description. See Cursor: Agent Skills and agentskills.io.

To verify locally:

python scripts/verify_github_install.py

If it prints OK: repo ready for Cursor/Claude/Codex (N skills)., the layout is correct. This check also runs in CI on every pull request. To confirm in Cursor: add the repo as a Remote Rule (Github), then open Settings → Rules and check that skills appear under Agent Decides.

Option B: Download release and copy

Skills live in skills/ (in the repo or in the release archive). Copy them into your environment’s skills directory.

Full installation details →

Summary:

  1. Download the latest release from the Releases page
  2. Extract the archive
  3. Copy skills/* to your environment’s skills directory (see table above: Cursor .cursor/skills/ or ~/.cursor/skills/, Claude .claude/skills/ or ~/.claude/skills/, Codex .codex/skills/ or ~/.codex/skills/). Cursor Team/Enterprise: Use Dashboard → Team Content → Skills to create team skills from skills/; they sync to your team.
  4. Restart your IDE or agent

Step 3: Run Your First Skill

Open your agent chat (e.g. Cursor: Cmd/Ctrl + L) and run:

/create-task --type=story for user profile page with avatar upload

AI will:

  1. Read the command instruction
  2. Connect to Jira via MCP
  3. Create story with title, acceptance criteria, and labels
  4. Return story ID (e.g., PROJ-123)

That's it! You're using agentic skills.

Try a Development Workflow

# Plan
/create-plan for PROJ-123

# Build
/start-task PROJ-123

# Ship
/complete-task PROJ-123

How It Works

AI agents handle routine tasks:

  • Story and epic creation
  • Implementation planning
  • Branch and PR creation
  • Test generation
  • Issue updates

You control decisions:

  • Plan approval
  • Code review
  • Work selection
  • Architecture

Full transparency:

  • Specs in specs/ folder (permanent feature contracts)
  • Plans in .plans/ folder (transient task implementation)
  • PRs link to stories
  • Commits include story IDs

See the product flow diagram →

Next Steps

Troubleshooting

Skills don't appear:

  • Verify skill folders in your environment’s skills directory (e.g. Cursor: ~/.cursor/skills/ or .cursor/skills/)
  • Restart your IDE or agent completely

MCP authorization fails:

  • Check Settings → MCP shows "Connected"
  • Verify GitHub token has repo and read:org scopes
  • Re-authorize Atlassian MCP

Skill fails:

  • Check error message (AI explains the issue)
  • Verify story/repo exists and you have permissions
  • Ensure you're in a git repository

Integration error:

  • Verify credentials and permissions in MCP settings.

Plan not found:

  • Run /create-plan before /start-task.

Story missing detail:

  • Commands will ask for clarification when information is insufficient.