Project Setup

Project & Agent Setup Workflow

This skill is used to initialize or update a project's settings, define stakeholders for automated agent notifications/@mentions, configure environment variables (.env), and generate matching MCP server configurations for the user's chosen agentic CLI.

Trigger

Use this skill when:

• Setting up a new repository with this agentic ruleset.
• The user asks to "configure the project", "setup the agents", "run project setup", or "setup mcp servers".
• An agent detects that both the package.json "agents" block and docs/project.json are missing or incomplete during startup checks.

---

Phase 1: Stakeholder & Project Configuration (package.json or docs/project.json)

The single source of truth for repository details and stakeholder GitHub usernames for mentions and reviews is stored in the package.json file under the "agents" configuration key. If package.json is missing or uninitialized, the workflow automatically falls back to docs/project.json.

1.1 Checked Configuration Schema

Every agent session must verify if package.json (under "agents") or docs/project.json exists. The expected schema structure is:

{
  "projectName": "your-project-name",
  "organization": "your-github-org",
  "repository": "your-github-repo",
  "stakeholders": {
    "productManager": "@pm-github-username",
    "seniorDeveloper": "@senior-dev-username",
    "qaLead": "@qa-lead-username",
    "techArchitect": "@architect-username"
  },
  "environments": {
    "production": "https://your-domain.com",
    "development": "https://dev.your-domain.com",
    "localPort": 6767
  }
}

1.2 Interactive Setup Script

To easily create or update this file without manual JSON editing, run the bundled interactive setup script:

npx node .agents/skills/project-setup/scripts/setup.js

---

Phase 2: Environment Credentials

During the setup process, the user is prompted for their API tokens. These are written directly to .env or .env.local to enable seamless integration with Figma and GitHub REST APIs:

• GITHUB_TOKEN: Personal access token with repo and workflow scopes (required for gh CLI and git automation).
• FIGMA_TOKEN: Personal access token (required for Figma Developer MCP to parse frames).

---

Phase 3: MCP Server Configuration

Depending on the user's preferred agentic CLI, the setup script generates and places the correct MCP server configuration file:

3.1 Antigravity CLI (agy) & Gemini CLI

• Location: .agents/mcp_config.json
• Stores custom tools like chrome-devtools, figma-developer-mcp, and github-mcp.

3.2 Cline (VS Code Extension)

• Location: .vscode/cline_mcp_settings.json
• Configures local node/npm tools to be visible within Cline's active session.

3.3 Qwen CLI

• Location: .qwen/settings.json
• Configures workspace settings, permissions, Gemini model providers, and MCP servers.

3.4 OpenCode

• Location: .opencode/settings.json
• Configures workspace settings, permissions, Gemini model providers, and MCP servers.

---

Phase 4: Verification

At the end of the setup flow, the setup script:

1. Validates that the configuration database (package.json or project.json) is well-formed.
2. Checks if gh auth status is authenticated (if a token was provided).
3. Verifies that the .env file contains the required variables.
4. Outputs instructions on how to load/reload MCP servers in the user's chosen CLI.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: P2ERGmbH
• Source: P2ERGmbH/agentic-coding
• License: MIT

Install and usage instructions live in the source repository linked above.