How to Use Chipp with Claude Code

The Chipp MCP Server brings the full power of the Chipp platform directly into Claude Code. Instead of switching between your terminal and the Chipp dashboard, you can create apps, manage knowledge sources, configure custom actions, and analyze conversations—all through natural language commands.

This guide walks you through everything you need to know to become productive with Chipp MCP in Claude Code.

The Big Picture

Here's what happens when you use Chipp MCP with Claude Code:

1. You speak naturally — Ask Claude to create an app, add documents, or check analytics
2. Claude calls the right tools — The MCP server translates your request into API calls
3. Chipp does the work — Apps are created, files are indexed, settings are updated
4. You see the results — Claude confirms what happened and suggests next steps

No clicking through menus. No copy-pasting IDs. Just describe what you want.

Getting Started

Prerequisites

Before you begin, make sure you have:

• Claude Code installed and running
• A Chipp account (Free tier works, but some features require Pro or Team)
• Your Chipp API key from the dashboard

Connecting the MCP Server

Follow our detailed setup guide to connect the Chipp MCP server to Claude Code:

Chipp MCP Server Setup Guide

The guide covers:

• Getting your API key from the Chipp dashboard
• Configuring Claude Code (and other MCP clients like Cursor)
• Verifying the connection is working

Once connected, Claude Code gains access to 80+ tools for managing your Chipp apps.

Core Capabilities

1. App Management

Create, update, and configure AI applications without leaving your terminal.

What you can do:

• Create new apps with custom system prompts
• Update app settings, models, and branding
• Search across all your apps
• Delete apps you no longer need

Example prompts:

"Create a customer support app for my SaaS product"
"Change my FAQ bot to use Claude Sonnet 4"
"List all apps in my workspace"
"Update the system prompt for app 12345 to be more conversational"

2. Knowledge Sources

Add documents, websites, and videos to give your apps domain expertise.

Supported source types:

• Documents: PDF, DOCX, TXT, MD, PPTX, XLSX, CSV, JSON, HTML
• Web content: Any URL, with optional full-site crawling
• Video: YouTube videos (transcripts are extracted automatically)
• Social: Instagram, TikTok, Facebook posts

Example prompts:

"Add this PDF to my support bot's knowledge base"
"Crawl the entire docs.example.com site for my app"
"Upload these three documents to app 456"
"What knowledge sources does my FAQ bot have?"

How document processing works:

1. Upload — You provide a file or URL
2. Extraction — LlamaParse extracts text from complex documents
3. Chunking — Content is split into semantic segments
4. Embedding — Each chunk gets a vector representation
5. Indexing — Vectors are stored for fast retrieval

Pro Tip: Use extractionInstructions when uploading documents to guide the parser. For example: "Focus on the FAQ section and ignore the legal disclaimers."

3. Custom Actions (API Integrations)

Connect your apps to external services so they can take real actions.

What custom actions enable:

• Query your database or CRM
• Create tickets in your helpdesk
• Send notifications via Slack or email
• Fetch real-time data from any API

Example prompts:

"Create an action that searches our product catalog"
"Add a HubSpot integration to look up contacts"
"Test the 'create ticket' action with sample data"
"List all custom actions on my support bot"

Parameter types:

4. Conversation Analytics

Understand how users interact with your apps.

Metrics you can access:

• Total conversations and messages
• Unique users over time
• Messages by source (web, API, Slack, WhatsApp, email)
• Top queries and common topics
• Model usage breakdown

Example prompts:

"Show me analytics for my FAQ bot over the last 30 days"
"What are the top 20 questions users ask my support bot?"
"Export all conversations from last week"
"Which of my apps gets the most traffic?"

5. Evaluations (Evals)

Test your apps against expected responses to catch regressions.

Note: Evals require a Business plan subscription.

How evals work:

1. Create a baseline — Start from an existing chat session
2. Define expectations — Edit responses to show what's correct
3. Run the eval — Claude re-answers each question
4. Compare results — See pass/fail with detailed scoring

Evaluation methods:

Example prompts:

"Create an eval for my FAQ bot based on session abc123"
"Run all evals for app 456"
"Show me the results of the last eval run"

6. Message Tags (Intent Detection)

Automatically detect user intents and trigger actions.

Use cases:

• Flag messages requesting human support
• Detect frustrated or angry users
• Identify upsell opportunities
• Route conversations by topic

Example prompts:

"Create a tag for 'wants to cancel subscription' with 50 training phrases"
"Generate utterances for detecting frustrated users"
"List all tags on my support bot"
"Show me messages that matched the 'billing issue' tag"

Pro Tip: Use generate_utterances to create diverse training phrases. More utterances = better detection accuracy.

7. Deployment Channels

Your apps can live anywhere your users are.

Supported channels:

• Web embed — Add a chat widget to any site
• Slack — Deploy as a Slack bot
• WhatsApp — Connect via WhatsApp Business
• Email — Auto-respond to incoming emails
• Voice — Phone calls via Twilio/LiveKit
• API — Build custom integrations

Example prompts:

"What Slack workspaces is my app connected to?"
"Show voice configuration for app 789"
"Get email deployment settings for my support bot"

8. AI-Powered Helpers

Let Claude help you build better apps.

Available AI tools:

Example prompts:

"Generate a system prompt for a legal assistant chatbot"
"Improve my FAQ bot's prompt to be more friendly"
"Suggest conversation starters for my onboarding app"
"What model should I use for a coding assistant?"

Real-World Workflow

Let's walk through building a complete customer support bot:

Step 1: Create the App

"Create an app called 'Acme Support Bot' in my main workspace.
It should be friendly and helpful, answering questions about
our software product."

Claude creates the app and returns the ID.

Step 2: Add Knowledge

"Add our product documentation from docs.acme.com to the support bot.
Crawl all linked pages."

Claude starts the crawl and monitors progress.

Step 3: Configure Actions

"Add a custom action that creates support tickets in our Zendesk
instance when users need human help."

Claude sets up the API integration with proper parameters.

Step 4: Set Up Intent Detection

"Create a tag to detect when users are frustrated or angry.
Generate 100 training utterances."

Claude creates the tag and populates it with diverse phrases.

Step 5: Test the App

"Chat with my support bot and ask 'How do I reset my password?'"

Claude tests the app and shows you the response.

Step 6: Create Evals

"Create an eval based on that conversation. The answer should
mention going to Settings > Security > Reset Password."

Claude sets up automated testing for quality assurance.

Note: This step requires a Business plan subscription.

Step 7: Check Analytics

"Show me how the support bot is performing this week."

Claude pulls metrics on conversations, top queries, and user engagement.

Best Practices

For Knowledge Sources

• Be specific with extraction instructions — Tell the parser what to focus on
• Refresh URLs periodically — Keep web content up to date
• Monitor processing status — Check that documents completed successfully
• Use hybrid search — Combine keyword and semantic matching for best results

For Custom Actions

• Test before deploying — Use test_custom_action to verify API calls work
• Use variables for secrets — Store API keys in app variables, not action definitions
• Chain actions carefully — Use TOOL_OUTPUT to pass data between actions
• Add clear descriptions — Help the AI know when to use each action

For System Prompts

• Start with generation — Let generate_system_prompt create a solid foundation
• Iterate with improvements — Use improve_system_prompt for refinements
• Be specific about tone — Define how the AI should communicate
• Include examples — Show the AI what good responses look like

For Evals

• Test critical paths — Focus on your most important use cases
• Run after changes — Check that updates don't break existing behavior
• Review failures carefully — Failed evals reveal prompt or knowledge gaps
• Use terminal format — Pass format: "terminal" for readable output

Quick Reference

Most-Used Commands

Tool Categories

Troubleshooting

"App not found"

Make sure you're using the correct app ID. Use list_apps to see all available apps.

"Knowledge source still processing"

Document processing takes time. Use get_knowledge_source to check status, or get_crawl_progress for URL sources.

"Custom action failed"

Use test_custom_action with parameterOverrides to debug. Check that your API endpoint is accessible and credentials are correct.

"Eval results seem wrong"

Review the expected responses in your eval. The AI compares against these, so they need to be accurate examples.

Continue Reading

• How Knowledge Sources Work — Deep dive into RAG and embeddings
• Custom Actions Overview — Advanced API integration patterns
• Analytics & Chat History — Understanding your app's performance
• Chipp MCP Server Setup — Connection and configuration details