- Home
- Skills
- Agents & Orchestration
- before anything
before anything
Professional engineering lifecycle for AI-driven projects, from discovery interviews to repo blueprinting.
Free
Works with the AI Agents you already use
before anything
Example session with this skill installed
I want to build a simple booking app for my local yoga studio. I have no code yet. Start the process.
- Read your context and instructions
- Compiled the before anything
I will start by gathering the necessary context to build the right architecture. I am loading the discovery-interview and stack-selection skills to ensure we don't build a system that is too complex or insecure for your needs.
Project Discovery: Yoga Studio Booking
I hear: A booking system for a yoga studio where students can view classes and reserve a spot. Before I scaffold the repo, I have a few questions to help me pick the right tools.
| Category | Question | Default Choice (Say "you pick") |
|---|---|---|
| WHO | Who manages the schedule, and how tech-savvy are they? | The studio owner, using a laptop. |
| JOB | What is the one thing a student MUST be able to do? | See a list of classes and book a spot. |
| SCALE | How many students do you have, and how many book at once? | 100 students, steady growth. |
| STAKES | If the site goes down for 3 hours, how bad is it? | Annoying, but not a business failure. |
| MONEY | Do students pay in the app or at the door? | Pay at the door (simple). |
| DATA | Do you need to store health info or just names/emails? | Names and emails only. |
| REALITY | When does this need to be live? | Within two weeks. |
Next steps
- Answer the questions above or say "you pick" for the defaults.
- I will then provide a Decision Record for your tech stack (e.g., Next.js + Supabase vs. a simple static site).
- Once the stack is confirmed, I will generate the
docs/BRIEF.mdand initial folder structure.
Connects securely to your tools. The creator never sees your data.
About this skill
Building applications with AI often leads to "vibe coding" disasters: architecture built on assumptions, messy repos that become unreadable in weeks, and security holes from silent shortcuts. This skill package enforces a professional engineering lifecycle for AI-driven development. It bridges the gap between a founder's vision and a developer's requirements by translating plain-language goals into concrete technical structures.
What it does
- Requirement gathering via discovery interviews that ask about users and stakes instead of SQL or NoSQL.
- Stack selection based on project scale and maintenance needs rather than whatever the model saw most recently.
- Repository blueprinting to establish naming conventions, folder structures, and essential documentation from day one.
- Workflow orchestration that forces a strict order of operations: understand, choose, design, build, then verify.
- Communication guardrails that ensure the agent uses plain language and never claim a test passed unless it actually ran.
How it works
- Initialize the orchestrator at the start of any new project or major feature request to score the task and load relevant sub-skills.
- Conduct the discovery interview using seven blocks (Who, Job, Scale, Stakes, Money, Data, Reality) to define the project boundaries.
- Generate the project blueprint to scaffold the
docs/,app/,lib/, andtests/directories with standardized naming rules.
Frameworks & tools
Works with any stack including Next.js, Remix, SvelteKit, FastAPI, Express, and mobile frameworks like Expo or Flutter.
Why this beats prompting it yourself
Standard prompts often produce one-shot code that lacks long-term maintainability. This skill enforces a multi-step engineering process that prevents the "AI-generated mess" by requiring architectural decisions before a single line of logic is written.
Use cases
- Starting a new SaaS MVP where you need to choose the right database and auth provider quickly.
- Organizing a sprawling codebase where files are piling up in the root directory.
- Handing off an AI-built project to a human developer or another agent.
Known limitations
Does not provide emergency incident response for production outages. Requires manual approval for installing new dependencies or spending money.
Not for emergency production fixes. Requires manual confirmation for paid services or new dependencies. Best suited for new projects or significant feature additions.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
1 use
Times used to date
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 27 days ago
- Free with an Agensi account
Requires