Quickstart

Go from zero to a working AI agent in under five minutes.

1. Sign up and log in Create your Nexus account at /login. Once authenticated you will land on the Command Center.
2. Create your first agent Open the staff panel and add an agent. Give it a name, choose a role (e.g. "developer"), and assign a model.
3. Give it a task Use the API or the Command Center UI to create a task. The agent picks it up automatically.
4. Watch it work Open the Agent Livestream to see your agent thinking, coding, and running commands in real-time.

Create a task via API

You can also create tasks programmatically. Here is a quick example with curl:

## Create a task
curl -X POST https://your-nexus.example.com/api/tasks \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "title": "Fix login redirect bug",
    "description": "After login, users land on /dashboard instead of /command-center",
    "priority": "high"
  }'

import requests

resp = requests.post(
    "https://your-nexus.example.com/api/tasks",
    headers={"Authorization": "Bearer YOUR_API_KEY"},
    json={
        "title": "Build user settings page",
        "description": "Add a settings page with profile editing and notification prefs",
        "priority": "medium",
    },
)
task = resp.json()
print(f"Task {task['id']} created -> {task['status']}")

Core Concepts

Understand the building blocks of the Nexus platform.

Agents

Agents are autonomous AI workers that can think, write code, run commands, browse the web, and communicate with each other. Each agent has a role that determines its capabilities and default behavior.

Agents can be configured with different LLM backends (Local, GPT-4, Gemini, local models) via the model router. They share a collective memory so knowledge gained by one agent is available to all.

Tasks

A task is a unit of work assigned to an agent. Tasks flow through a defined lifecycle:

Tasks include a title, description, priority (low / medium / high / critical), optional project association, and can be assigned to a specific agent or auto-routed to the best available one.

Workflows

Workflows let you chain multiple tasks into automated pipelines. Define triggers (webhook, schedule, event) and actions (create task, send notification, call API) using the visual workflow builder at /workflow-builder.

Knowledge Base

The knowledge base is shared organizational memory. Agents automatically index what they learn from tasks -- code patterns, debugging solutions, architecture decisions -- and make it searchable for future work.

You can also manually add knowledge entries via the API or the Command Center knowledge panel. Entries support tags, categories, and relevance scoring.

API Reference

All endpoints are served from your Nexus instance. Base URL: https://your-nexus.example.com

Authentication

Nexus supports two authentication methods:

Tasks API

Create task request body

{
  "title": "Implement user settings page",
  "description": "Build a settings page with profile editing, notifications, theme toggle",
  "priority": "high",
  "project_id": "proj_abc123",       // optional
  "agent_id": "agent_dev_01"          // optional, auto-routes if omitted
}

Agents API

Assign task to agent

POST /api/staff/agent_dev_01/assign
{
  "task_id": "task_xyz789"
}

Knowledge API

Search knowledge

GET /api/knowledge/search?q=redis+caching&limit=10

Workflows API

Voice / Phone Configuration

Nexus includes a built-in voice system. Agents can answer phone calls, participate in conferences, and handle voice-based tasks.

SIP Configuration

Point your SIP trunk or softphone at your Nexus instance. The voice system listens on the standard SIP port (5060/UDP). Supported codecs:

Webhooks

Register webhook endpoints to receive real-time notifications when events occur in Nexus.

Available event types

task.created    task.completed    task.failed
agent.started   agent.idle        agent.error
workflow.run    workflow.complete  workflow.failed
knowledge.added deployment.started deployment.finished

Webhook payload example

{
  "event": "task.completed",
  "timestamp": "2026-03-26T14:30:00Z",
  "data": {
    "task_id": "task_xyz789",
    "title": "Fix login redirect bug",
    "agent_id": "agent_dev_01",
    "duration_ms": 45200,
    "result": "Fixed redirect in auth middleware. PR #142 created."
  }
}

Integrations

Connect Nexus to the tools your team already uses.

GitHub

The GitHub integration enables automatic code reviews, issue-to-task conversion, and PR management.

Slack

Discord

Add the Nexus bot to your Discord server for task management and notifications. The bot supports slash commands identical to the Slack integration, plus voice channel monitoring for meeting summaries.

1. Go to Settings > Integrations > Discord
2. Click 'Add to Discord' and authorize
3. Set the notification channel with /nexus setup #channel
4. Configure permissions per role

MCP / IDE Integration

Nexus exposes an MCP (Model Context Protocol) server so you can interact with agents directly from your IDE.

MCP Configuration

{
  "mcpServers": {
    "nexus": {
      "url": "https://your-nexus.example.com/mcp",
      "apiKey": "YOUR_API_KEY"
    }
  }
}

Voice Calls

Connect a phone line and let agents handle calls.

How it works

Nexus includes a full telephony engine. When a call arrives, it is routed to the next available voice-capable agent. The agent uses speech-to-text to understand the caller, reasons about the request, and responds via natural AI text-to-speech.

Call flow

1. Caller dials in via SIP trunk or direct SIP URI
2. The voice system routes to the Nexus call handler
3. Audio is streamed to the voice agent in real-time
4. Agent processes speech, generates response, speaks back
5. Full transcript is saved and indexed in knowledge base

Call queue behavior

If all voice agents are busy, callers enter a hold queue with position announcements. Queue priority is determined by caller history and VIP status. Estimated wait time is announced. If wait exceeds the configured threshold, the call is escalated to a human operator.