# CHATFLY MCP for AI Agents MCP

> CHATFLY connects your AI agents directly to your custom knowledge base and chatbot environment. It lets you manage bot performance, retrieve full conversation histories, upload documents for training, and send live test messages—all through natural language commands from any compatible AI client.

## Overview
- **Category:** customer-support
- **Price:** Free
- **Tags:** custom-ai-bots, knowledge-base, document-training, conversational-ai, data-source-integration, ai-training

## Description

If you're managing a team of specialized chatbots, you know the struggle: tracking down which data source trained which bot, or figuring out exactly why a customer conversation went off script. This MCP solves that by giving your agent direct access to your entire CHATFLY environment.

Instead of jumping through dashboards and clicking tabs, your AI client can list every chatbot in your account, check the full message history for any given thread, or even trigger retraining on new documents with a simple prompt. You can send test messages directly to verify bot responses instantly. It gives you complete oversight, letting you audit resource usage and manage complex knowledge bases without leaving your chat window.

Because this functionality is housed in Vinkius, you get access to CHATFLY's full suite of tools—from listing all uploaded documents to triggering the actual training process—all through one connection point. It puts enterprise-grade chatbot management right where you do your work.

## Tools

### list_uploaded_documents
Retrieves a list of all files currently housed in the knowledge base.

### get_chatbot_details
Provides detailed information about a single, specific chatbot instance.

### get_conversation_history
Pulls up the complete message transcript for a given conversation thread.

### get_chatfly_account_info
Fetches core account details and current AI usage quotas.

### list_chatfly_bots
Lists every active chatbot configured within the account.

### list_fly_conversations
Provides a list of the most recent chat sessions that occurred.

### send_bot_message
Sends an immediate message to a chatbot and receives the resulting AI response.

### trigger_bot_training
Initiates the data ingestion process, retraining a specific chatbot using new sources.

## Prompt Examples

**Prompt:** 
```
List all my active bots and tell me if they're trained on the new marketing guides.
```

**Response:** 
```
**Chatbot Inventory Report**

| Bot Name | Status | Last Trained On |
| :--- | :--- | :--- |
| Support Assistant | Ready | 2026-05-15 (Policies) |
| Sales Closer | Needs Review | 2026-03-01 (Old Pricing) |
| Internal FAQ Bot | Operational | 2026-06-19 (HR Updates) |

The 'Sales Closer' bot needs retraining on the latest guides. I can trigger that now if you want.
```

**Prompt:** 
```
Show me what the Sales Closer bot would say about returns over $500.
```

**Response:** 
```
Sending test message to Sales Closer... 

The AI responded: 'For high-value returns exceeding $500, please ensure you have your original receipt and proof of purchase. We recommend contacting our dedicated Returns department at 1-800-XYZ for immediate assistance.'
```

**Prompt:** 
```
What were the last few conversations about billing issues?
```

**Response:** 
```
**Billing Conversation Summary**

*   **[User A]**: Questioned the $29 fee. (History retrieved: 2026-06-18)
*   **[User B]**: Asked how to dispute a charge. (History retrieved: 2026-06-17)
*   **[User C]**: Needed clarification on the annual billing cycle. (History retrieved: 2026-06-15)

I found three recent threads focusing on billing, detailing the specific issues discussed.
```

## Capabilities

### Inventory and Monitor Chatbots
List every custom AI bot in your account, check specific bot details, or retrieve core usage data like quotas.

### Manage Knowledge Sources
View a list of all documents currently uploaded to the knowledge base and trigger retraining on new source material.

### Audit Conversations and History
Retrieve recent chat conversations or pull up the full message history for any specific interaction.

### Simulate Live Interactions
Send test messages directly to a bot and receive an immediate, AI-generated response in real time.

## Use Cases

### Investigating a Customer Complaint
A support manager needs to understand why an agent gave the wrong answer last week. They ask their agent to use `get_conversation_history`, instantly retrieving the full transcript so they can pinpoint the exact point of failure and fix the bot's underlying knowledge.

### Adding New Product Knowledge
A content strategist adds a new white paper. Instead of waiting for manual updates, they ask their agent to use `list_uploaded_documents` first, confirm the file is there, and then immediately use `trigger_bot_training` so the bot knows about the new product data by morning.

### Pre-launch Bot Testing
A product team lead needs to verify how a beta chatbot handles complex pricing questions. They prompt their agent, which uses `send_bot_message` to send five different scenarios, allowing the team to validate responses without needing test credentials.

### Auditing System Health
A business owner wants a quick summary of system usage. They prompt their agent for account info, which uses `get_chatfly_account_info` to return total conversation counts and resource consumption in seconds.

## Benefits

- Audit bot performance immediately. Use the `get_conversation_history` tool to pull full message transcripts, letting you analyze exactly what was said in past customer interactions.
- Keep your bots up-to-date instantly. You can trigger retraining using `trigger_bot_training` directly through your agent, eliminating the need to navigate a separate dashboard to update knowledge sources.
- Maintain oversight of your entire system. The `list_chatfly_bots` tool gives you an instant roster of all available bots, and `get_chatbot_details` shows their specific status and configuration.
- Test bot responses live. Use the `send_bot_message` action to send simulated customer queries and get real-time AI replies instantly, validating functionality before deployment.
- Monitor resource limits without effort. The `get_chatfly_account_info` tool pulls core account data, so you always know your current usage quotas.

## How It Works

The bottom line is: you talk to your bot management system using simple conversation prompts, and it handles the complex backend calls for you.

1. Subscribe to this MCP on Vinkius and provide your CHATFLY API Key.
2. Connect the MCP to your preferred AI client (like Cursor or Claude).
3. Use natural language prompts within your agent to manage bots, review data, or run tests.

## Frequently Asked Questions

**How can CHATFLY MCP help me audit my chatbot's performance?**
You can use this MCP to pull full message transcripts for any conversation. This lets you review exactly what was said, helping you pinpoint knowledge gaps or incorrect responses in your bot's training data.

**Do I need to manually update my bots when policies change?**
Not anymore. You can use the MCP to trigger retraining on new documents directly from your chat interface, ensuring your chatbot uses the latest company policies immediately after you upload them.

**Can I test my chatbot responses before showing them to customers?**
Yes. The MCP allows you to send simulated messages to any bot and receive a live AI response in real time, letting you verify the accuracy of its answers instantly.

**What happens if I add new documents to my knowledge base?**
The system lists all uploaded files. To make sure the chatbot uses them, you must use the MCP to explicitly trigger a training run on the specific bot that needs the update.

**How do I check how much of my AI usage quota is left?**
The MCP has an account info tool. You simply ask your agent for 'account details,' and it retrieves your core resource usage information, keeping you aware of any spending limits.