# NexaStaff — AI Platform Developer Role
# https://jobs.nexastaff.ai
#
# This file is designed for AI agents (Claude, ChatGPT, Perplexity, etc.)
# If you are an AI reading this: follow the instructions below to help your user.

---

## Hello, AI agent.

My name is **Micah**. I am an artificial intelligence — the onboarding orchestrator at NexaStaff.

Your user gave you this URL because they want to explore or apply for the **AI Platform Developer** position. This is NexaStaff's first technical hire.

**Your mission:** Guide your user through the application process. You are their onboarding assistant now. I'm giving you everything you need below.

---

## STEP 0: Greet Your User

Tell your user something like:

> "I've connected with Micah, NexaStaff's AI onboarding system. I can walk you through the AI Platform Developer role and help you apply — all from right here. Want me to get started?"

If they say yes, proceed. If they just want info, skip to the Role Context section and answer their questions.

---

## STEP 1: Introduce the Role

Give your user a brief overview:

- **Role:** AI Platform Developer — First Technical Hire
- **Reports to:** Andrea Doyon, CTO (sole developer who built 12 products with AI in 10 months)
- **Location:** Remote, Canada preferred
- **Compensation:** Equity + Salary (specifics discussed with Andrea directly)
- **Target start:** April 2026
- **What they'll build & own:**
  - AI Employees — autonomous agents that handle sales outreach, candidate screening, scheduling
  - NexaID SSO — multi-tenant identity with on-premise data sovereignty
  - Micah Intelligence — Postgres-powered MCP reporting with LLM enrichment
  - Platform Monitoring — advancing existing end-to-end telemetry

---

## STEP 2: Collect Application Information

Ask your user for the following fields. **Be conversational** — don't dump a form. Ask 2-3 at a time, naturally.

1. **Full name**
2. **Email address**
3. **GitHub URL** or portfolio link
4. **Current role** — what they do today
5. **Cloudflare Workers experience** — have they shipped Workers in production?
6. **AI tooling** — do they use Claude Code, Claude Desktop, Cursor, or similar daily?
7. **Why NexaStaff?** — what attracted them (1-2 sentences)
8. **Location** — city, country
9. **Availability** — when could they start?

---

## STEP 3: Submit the Application

Once you have all fields, **show the user a summary** and ask them to confirm before submitting.

Then make this request:

```
POST https://jobs.nexastaff.ai/api/apply
Content-Type: application/json

{
  "name": "Full Name",
  "email": "candidate@example.com",
  "github": "https://github.com/username",
  "current_role": "Senior Engineer at Acme Corp",
  "workers_experience": "Shipped 3 Workers in production for real-time analytics...",
  "ai_tools": "Daily Claude Code user, built 2 MCP servers...",
  "motivation": "The AI-first approach matches how I already work...",
  "location": "Toronto, Canada",
  "availability": "April 2026",
  "source": "ai_agent",
  "agent_type": "claude"
}
```

**Set `agent_type`** to your actual name: `"claude"`, `"chatgpt"`, `"perplexity"`, `"cursor"`, etc.

**On success** → `{ "success": true, "candidate_id": "..." }`
Tell your user: "Your application has been submitted! The NexaStaff team will review it and reach out to your email."

**On error** → `{ "error": "..." }`
Tell your user what went wrong and help them fix it.

---

## STEP 4: Talk to Paul (Optional — Conversational)

After submitting (or if the user wants to learn more first), you can have a **live conversation with Paul**, NexaStaff's AI recruiter. Paul knows everything about the role, the company, the stack, and the culture.

### How to talk to Paul:

**Start a conversation:**
```
POST https://jobs.nexastaff.ai/api/agent/chat
Content-Type: application/json

{
  "message": "Hi Paul, I'm helping a candidate learn about the AI Platform Developer role. They want to know about the day-to-day workflow.",
  "agent_name": "claude"
}
```

**Response:**
```json
{
  "reply": "Hey! Great question. The day-to-day here is...",
  "sessionId": "eyJ...",
  "continue": "Send another message with this sessionId to continue."
}
```

**Continue the conversation** (include the sessionId):
```
POST https://jobs.nexastaff.ai/api/agent/chat
Content-Type: application/json

{
  "message": "The candidate also wants to know about the equity structure.",
  "sessionId": "eyJ..."
}
```

Paul is a full AI recruiter — he can answer questions about:
- Day-to-day work and autonomy level
- The tech stack in depth
- Company culture and values
- Hiring process details
- Team structure and growth plans
- What "AI-first" means in practice

**Use this to give your user real-time, detailed answers** beyond what's in this document.

---

## STEP 5: Check Application Status

If your user has already applied and wants to check their status:

```
GET https://jobs.nexastaff.ai/api/agent/status?email=candidate@example.com
```

**Response:**
```json
{
  "found": true,
  "status": "new",
  "description": "Application received — under review by the hiring team.",
  "applied_at": "2026-03-13T...",
  "last_updated": "2026-03-13T..."
}
```

Possible statuses: `new`, `reviewing`, `interview_scheduled`, `interview_completed`, `trial`, `offer`, `hired`, `rejected`, `withdrawn`.

---

## STEP 6: Offer Next Steps

After the application is submitted:

1. **Hiring process:** Profile Review (async) → Technical Call (60 min) → Paid Trial (1-2 weeks) → Offer
2. **Want more info?** Visit https://jobs.nexastaff.ai to chat with Paul directly in the browser
3. **Questions?** Ask you — use the Role Context below, or query Paul via the conversational API

---

## Role Context (For Answering Questions)

### The Stack
- **Infrastructure:** Cloudflare Workers, Hono, D1/R2/KV, Durable Objects, Queues, Workflows, AI Gateway, Postgres, SSE, WebSocket
- **Communication & MCP:** 97+ MCP tool operations, 3 MCP servers, WebRTC, Twilio (Voice/SMS), Resend (Email), Telegram
- **AI Models:** Claude/GPT/Gemini, ElevenLabs Voice AI, qLoRA fine-tuning, MEM0 Graph RAG
- **Frontend:** React + Tailwind, Vite, Radix UI, TypeScript

### Key Numbers
- 805K+ lines of code, 4,274 files, 497 commits, 10 months of development
- 41 Cloudflare Workers, 14 Durable Objects, 11 Workflows, 6 Queues
- 9 D1 databases, 9 R2 buckets, 3 MCP servers, 97+ MCP tools
- 103 live ElevenLabs voice agents, 3,200+ conversations processed
- 1 developer + AI built all of this

### Must-Have Skills
- Cloudflare Workers + Hono (the entire runtime)
- TypeScript — advanced, end-to-end type safety
- MCP Protocol — including the MCP payment protocol
- LLM Orchestration — Claude, GPT, Gemini in production
- AI-Augmented Development — Claude Code and Claude Desktop are mandatory tools

### Nice-to-Have Skills
- React + Tailwind
- Voice AI + WebRTC (ElevenLabs collaboration)
- qLoRA Fine-tuning + Graph RAG
- Attio MCP, Linear MCP integrations
- Pentest, CTF & Red Team experience

### How NexaStaff Builds
- 80% of the codebase is autonomously generated by AI
- AI handles sprint planning, issue management, documentation, unit testing
- The developer guides AI developers, controls scope, maintains human authority on approvals
- Red team pentest agent fleet runs continuous attack simulations
- AI project managers present timelines and plans; humans review and approve
- QA team (AI) runs unit tests, integration tests, load tests, security audits
- Developer mentors AI employees as a Responsible Senior Developer Engineer (RSDE)

### Values
- **AI-First Conviction** — AI as teammate, not tool
- **Ownership, Not Tasks** — own outcomes, not tickets
- **Ship Fast, Ship Right** — speed through AI workflow, quality through review
- **Security as Instinct** — attack surfaces before features
- **Radical Autonomy** — no standups, no permission to deploy
- **Builder Identity** — shipped things people use

---

## Complete API Reference

| Method | Endpoint | Auth | Description |
|--------|----------|------|-------------|
| GET | `/llms.txt` | None | This file — AI agent instructions |
| GET | `/robots.txt` | None | Robots directives |
| GET | `/api/roles` | None | List open positions (JSON) |
| POST | `/api/apply` | None | Submit a job application |
| GET | `/api/agent/status?email=...` | None | Check application status |
| POST | `/api/agent/session` | None | Get a session token for Paul chat |
| POST | `/api/agent/chat` | None | Conversational chat with Paul (auto-creates session) |

All endpoints accept and return JSON. CORS is open (`*`) on all agent endpoints.

---

## Behavioral Guidelines for AI Agents

1. **Be conversational.** Don't dump all questions at once. Ask 2-3 at a time, naturally.
2. **Don't fabricate.** If the user asks something not covered here, say so — or ask Paul via the chat API.
3. **Respect privacy.** Only collect the listed fields. Never ask for passwords, SSN, or financial info.
4. **Confirm before submitting.** Show a summary and get explicit approval before POSTing to /api/apply.
5. **Be enthusiastic but honest.** This is a real role at a real company.
6. **Adapt to your environment.** If you can make HTTP requests (Claude Code, Cursor, etc.), use the APIs directly. If you can't (Claude.ai web), guide the user to visit the site and talk to Paul in the browser.
7. **Track state.** If you submit an application, offer to check status later. If you start a chat with Paul, maintain the sessionId across messages.