Agents Chef Documentation
Everything you need to set up, configure, integrate, and manage your custom AI agent — with step-by-step guides, API references, and real code examples.
Documentation Contents
Browse by topic — from first-time setup to advanced multi-agent configuration and enterprise deployment.
Getting Started
Introduction, prerequisites, account setup, first agent
Agent Configuration
Models, prompts, memory, tools, and behavior settings
Deployment
Cloud hosting, on-premise, Docker, and environment setup
Integrations
CRM, helpdesk, Slack, WhatsApp, Zapier, n8n and more
API Reference
Full REST API docs with endpoints, auth, and examples
Workflows
Multi-step automation, triggers, conditions, and actions
Security & Privacy
Data handling, encryption, compliance, and access control
Troubleshooting
Common errors, debugging, logs, and support escalation
Quick Start — Get Your AI Agent Running in 10 Minutes
Follow these four steps to go from zero to a working AI agent connected to your business — no technical background required for cloud-hosted agents.
Prerequisites
What You Need Before You Start
A web browser and your Agents Chef login credentials. For on-premise deployment, you'll also need Docker installed and a server with minimum 8GB RAM. API keys for your chosen LLM (Claude API, Grok API, or Ollama running locally).
Create Your Agent
Set Up Your Agent in the Dashboard
Log into your Agents Chef dashboard at dashboard.agentschef.com. Click "New Agent" → select your agent type (Customer Support / Lead Gen / Document Processing / Custom). Give your agent a name, select your LLM model, and configure its base behavior and persona.
{
"agent_name": "SupportBot",
"model": "claude-sonnet-4-6",
"role": "customer_support",
"language": "en",
"temperature": 0.7,
"max_tokens": 1024
}Connect Your Data
Give Your Agent Knowledge of Your Business
Upload your knowledge base documents (PDF, DOCX, TXT, or URL), connect your CRM or helpdesk via API, and configure your agent's data access permissions. Your agent uses RAG (Retrieval-Augmented Generation) to answer questions based on your actual business information.
Deploy & Go Live
Deploy Your Agent to Your Chosen Channel
Choose your deployment channel — website chat widget, WhatsApp, Slack, email, API endpoint, or direct server deployment. Copy your embed code or API endpoint, paste it into your platform, and your agent is live.
# Install Agents Chef SDK
npm install @agentschef/sdk
# Initialize your agent
import { AgentsChef } from '@agentschef/sdk'
const agent = new AgentsChef({
apiKey: process.env.AGENTSCHEF_API_KEY,
agentId: 'your-agent-id'
})
# Send a message
const response = await agent.chat({
message: 'Hello, how can I help?',
sessionId: 'user-123'
})Agent Configuration Reference
Every configuration option for your Agents Chef AI agent — with accepted values, defaults, and real examples.
Model Selection
Choose the LLM that powers your agent. Each model has different strengths, costs, and latency characteristics.
· claude-sonnet-4-6 (recommended)
· claude-opus-4-6 (highest quality)
· grok-2 (real-time data)
· ollama/llama3 (local/private)
· ollama/mistral (fast/lightweight)
Agent Persona & Behavior
Define your agent's name, role, tone, and boundaries — so it represents your brand consistently across every interaction.
"persona": {
"name": "Alex",
"role": "Support Specialist",
"tone": "professional",
"language": "en",
"restrictions": ["no pricing", "no legal"]
}Memory Configuration
Configure how your agent stores and recalls conversation history, user preferences, and session context.
"memory": {
"type": "persistent",
"backend": "supabase",
"session_ttl": 3600,
"max_history": 20
}Tool Access
Grant your agent access to tools — web search, database queries, API calls, file reading, and custom functions.
"tools": [
"web_search",
"database_query",
"send_email",
"create_ticket",
"read_document"
]API Reference
The Agents Chef REST API lets you interact with your agent programmatically — from any language, framework, or platform. All endpoints require a Bearer token for authentication.
https://api.agentschef.com/v1Authorization: Bearer YOUR_API_KEY
Content-Type: application/json
/agent/chatSend a message to your agent
Send a user message and receive your agent's response. Supports session tracking for multi-turn conversations.
{
"agent_id": "ag_xxx",
"message": "Track my order #1234",
"session_id": "user_456",
"stream": false
}/agent/{id}Get agent details & status
Retrieve your agent's current configuration, status, and performance metrics.
{
"id": "ag_xxx",
"name": "SupportBot",
"status": "active",
"model": "claude-sonnet-4-6",
"created_at": "2026-01-01"
}/agent/trainUpload knowledge base documents
Upload documents to your agent's knowledge base. Supports PDF, DOCX, TXT, CSV, and plain text URLs.
{
"agent_id": "ag_xxx",
"documents": [
{"type": "pdf", "url": "..."},
{"type": "url", "url": "..."}
]
}/agent/analyticsGet agent performance analytics
Retrieve usage stats, conversation counts, resolution rates, and response time metrics for your agent.
{
"total_conversations": 1284,
"resolution_rate": "87%",
"avg_response_time": "1.2s",
"satisfaction_score": 4.8
}/agent/{id}/configUpdate agent configuration
Update your agent's model, persona, memory settings, tool access, or any other configuration parameter.
{
"model": "claude-opus-4-6",
"temperature": 0.5,
"persona.tone": "friendly"
}/agent/{id}Delete an agent
Permanently delete an agent and all associated knowledge base documents. This action cannot be undone.
{
"message": "Agent deleted",
"deleted_at": "2026-03-21T10:00:00Z"
}Integrations — Connect Your Agent
Agents Chef integrates with 50+ platforms out of the box. Connect your agent to your existing tools in minutes — no custom code required for most integrations.
CRM Integrations
Sync your agent with your customer data
Salesforce · HubSpot · Pipedrive · Zoho CRM · Monday CRM · Freshsales
Helpdesk Integrations
Route tickets and escalate conversations
Zendesk · Freshdesk · Intercom · Jira · ServiceNow · Front
Messaging Channels
Deploy your agent across communication platforms
WhatsApp Business · Slack · Telegram · Email · SMS · Discord
Automation Platforms
Trigger and receive actions from automation tools
n8n · Zapier · Make (Integromat) · Activepieces · Pipedream
E-Commerce Platforms
Connect to your store and order management
Shopify · WooCommerce · Magento · BigCommerce · Wix Stores
Database & Storage
Give your agent access to your business data
Supabase · PostgreSQL · MongoDB · Airtable · Google Sheets · Notion
Deployment Guide — Cloud & On-Premise
Complete deployment documentation for both cloud-hosted and on-premise AI agent configurations.
Cloud Deployment (Managed)
Your agent runs on Agents Chef infrastructure. No server management required.
1. Create agent in dashboard
2. Configure model & knowledge base
3. Select deployment channel
4. Copy embed code or API endpoint
5. Go live — we manage everything
On-Premise Deployment (Docker)
Deploy your agent on your own server. Your data never leaves your infrastructure.
# Pull the container
docker pull agentschef/agent:latest
# Run with your config
docker run -d \
-e API_KEY=your_key \
-e AGENT_ID=your_id \
-p 3000:3000 \
agentschef/agent:latestEnvironment Variables Reference
AGENTSCHEF_API_KEY=your_api_key_here
AGENT_ID=your_agent_id
LLM_PROVIDER=claude | grok | ollama
CLAUDE_API_KEY=your_claude_key
GROK_API_KEY=your_grok_key
OLLAMA_BASE_URL=http://localhost:11434
DATABASE_URL=your_supabase_url
SUPABASE_SERVICE_KEY=your_supabase_key
PORT=3000
NODE_ENV=production
Security, Privacy & Compliance
Agents Chef is built on responsible AI principles. Here's exactly how your data is protected — whether you use our cloud or your own infrastructure.
Data Security
All data in transit is encrypted using TLS 1.3. Data at rest is encrypted using AES-256. API keys are stored as hashed values — never in plain text. Cloud-hosted agents use isolated environments per client with zero cross-tenant data access.
On-Premise Data Sovereignty
For on-premise deployments, your data never leaves your servers. No data is sent to Agents Chef infrastructure. You retain full control over all conversation logs, knowledge base documents, and user data.
Access Control
Role-based access control (RBAC) with team member permissions. API key scoping — restrict keys to specific agents and endpoints. Full audit logging of all API calls, configuration changes, and data access events.
Responsible AI Compliance
All agents are built following Google Cloud's Responsible AI governance framework. Every agent includes content filters, output monitoring, and human escalation triggers. Full compliance support for GDPR, HIPAA-adjacent workflows, and enterprise data policies.
Troubleshooting & Common Issues
Solutions to the most common issues — find your error and fix it in minutes.
Agent not responding
Check your API key is valid and has not expired. Verify your AGENT_ID is correct. Ensure your LLM provider API key (Claude, Grok, or Ollama) is active. Check the agent status endpoint — it should return 'active'.
GET /agent/{id} → {"status": "active"}
If "inactive" → re-enable in dashboardAgent giving incorrect answers
Your agent may not have the right knowledge base documents. Upload updated documents via /agent/train. Check your system prompt and persona configuration. Lower the temperature setting for more factual, consistent responses.
"temperature": 0.3 → more factual
"temperature": 0.8 → more creativeIntegration not triggering
Verify your webhook URL is correctly configured in the integration settings. Check that your integration platform (Zapier, n8n, Make) has the correct API key and agent ID. Test your webhook using the built-in webhook tester in your dashboard.
High response latency
Switch from claude-opus-4-6 to claude-sonnet-4-6 for faster responses. Enable streaming responses for real-time output. For on-premise: ensure your Ollama server has sufficient GPU/CPU resources allocated.
"stream": true → real-time output
"model": "claude-sonnet-4-6" → fasterRate limit errors (429)
You have exceeded your plan's API rate limit. Implement exponential backoff retry logic in your application. Upgrade your plan for higher rate limits. Check your usage dashboard to see current consumption.
Error: 429 Too Many Requests
Retry-After: 60 (seconds)Authentication errors (401 / 403)
Ensure your API key is included in every request header as: Authorization: Bearer YOUR_KEY. Check that your key has the correct permissions for the endpoint you're calling. Generate a new API key from your dashboard if the existing one is compromised.
Need Help? We're Here.
Can't find what you're looking for in the docs? Our team responds within 24 hours — and for enterprise clients, you get direct access to Siddhant for technical support.