Agent-Startup-Skills

Usage Guide

This guide covers the core interaction patterns for the Agent Startup Skills ecosystem.

1. The Startup Factory (Main Workflow)

The saas-accelerator is your primary tool. It coordinates 8 specialized agents in a strictly governed sequence.

Activation: “I want to build a [Description of Idea] using the Startup Factory.”

The Approval Gate Mechanism

To prevent the agent from “going rogue,” every phase ends with a gate.

1. The agent produces a markdown artifact (e.g., architecture.md).
2. The agent pauses and asks for your approval.
3. You MUST type approve or approve phase to continue.

Feedback Loop: If the output isn’t perfect, provide feedback instead of approving: “Change the database to MongoDB and remove the social login feature.” The agent will regenerate the artifact and ask for approval again.

---

2. Standalone Skill Invocation

You don’t have to run the full pipeline. You can call any specialist at any time using their standalone trigger phrases or IDs.

Examples:

• @[market-scout] Research market for a peer-to-peer car sharing app.
• @[security-auditor] Audit the security of this project.
• @[system-architect] Design a system for a real-time chat app using WebSockets.

---

3. Customizing Your Tech Stack

By default, the agents prefer Next.js, TypeScript, and Tailwind. To change this, simply tell the Architect (Phase 3) or the Master Orchestrator:

“Build me a blog platform using the Startup Factory, but use Python with FastAPI and a PostgreSQL database.”

The agents will adapt all subsequent artifacts (Schema, Code, Tests, Deploy) to your specified stack.

This site is open source. Improve this page.