# Sift (Chargeback) MCP for AI Agents AI Agent Connect > Sift (Chargeback) lets you manage fraud prevention and chargeback events directly through your AI agent. You can check user fraud scores, report new disputes, and apply manual actions like blocking users or accepting orders without switching tabs. It connects your Sift account to any compatible client so you can handle risk analysis and trust and safety tasks using natural language. ## Overview - **Category:** security-compliance - **Price:** Free - **Endpoint:** https://edge.vinkius.com/vk_preview_X6jrj9G3NJ6qSu4scGacFQiORtOckhn98ppSzoMs/ai-agent-connect - **Tags:** fraud-prevention, chargeback-management, risk-analysis, dispute-resolution, fraud-detection, security-monitoring ## Description This Sift (Chargeback) MCP connects your account to any AI agent so you can handle fraud protection and chargeback management through natural conversation. Instead of jumping between tabs to hunt for a user ID or digging through logs to see if a transaction is risky, you can just ask your agent to check the score. You can see if a customer is flagged as a risk, look up their history of decisions, or report a new dispute in seconds. It takes the manual work out of risk analysis. Because Vinkius hosts this Connector, you can connect it to your existing workflow and handle things like blocking high-risk accounts or approving orders through a simple chat. It turns your AI into a teammate that actually knows your security rules and can act on them immediately. You can also see your active security rules and log custom events like logins to help improve your security model over time. This means you spend less time clicking and more time actually managing your business safety. You get a clear picture of your risk landscape without the headache of navigating a complex dashboard. You can monitor your fraud prevention workflows in real time and see exactly what labels have been applied to your users. It makes the whole process of trust and safety management feel like a conversation rather than a chore. ## Tools ### apply_user_decision Apply a manual decision like blocking a user to your account. This lets you act on risk immediately. ### list_user_decision_history See a full list of past decisions for a specific user. It is great for auditing account history. ### get_user_fraud_labels See the specific labels like good or bad assigned to a user. This helps you understand their risk status. ### get_user_fraud_score Get the current fraud score for any user. You can use this to decide if a transaction is safe. ### list_sift_decisions See all the different actions you can take in Sift. This helps you know what options are available. ### list_sift_workflows See your configured fraud prevention workflows. This gives you a clear view of your security rules. ### report_sift_chargeback Report a new chargeback event to Sift. It updates the user's risk profile automatically. ### track_sift_event Log custom events like logins or transactions. This helps refine the machine learning model. ## Prompt Examples **Prompt:** ``` What is the risk for user 'user_abc_123'? ``` **Response:** ``` Checking risk for 'user_abc_123'... The user has a fraud score of **85.2% (High Risk)**. Sift has labeled this user as **$bad** based on recent transaction patterns. I recommend reviewing their recent activity before approving any further orders. ``` **Prompt:** ``` Report a chargeback for order #999 as fraud. ``` **Response:** ``` Reporting chargeback for order #999... **Success:** The event has been sent to Sift with the reason **$fraud**. This will update the user's risk profile immediately to help prevent future issues. ``` **Prompt:** ``` What actions were taken on 'user_456'? ``` **Response:** ``` Retrieving decision history for 'user_456'... I found the following recent decisions: - **watch_user** (Applied on 2024-03-15) - **approve_order** (Applied on 2024-02-10) No other recent actions were found. ``` ## Capabilities ### Check user fraud scores Get real-time risk ratings for any customer to see if they are likely to be fraudulent. ### Report chargeback events Notify Sift about new disputes and their reasons to keep your risk profile current. ### Apply user decisions Block users or accept orders with a single command to manage risk instantly. ### Audit decision history See a full list of past decisions and actions taken on a specific account. ### View fraud labels See if a user is currently marked as good or bad based on Sift's logic. ### Monitor fraud workflows See your active security rules and configurations in real time. ### Log custom events Track logins or transactions to help refine your fraud prevention model. ## Use Cases ### High-risk transaction alert A manager sees a $5,000 order and asks the agent to check the fraud score. The agent sees it is 90% risk and warns the manager before they ship the item. ### Handling a refund dispute A support rep needs to report a chargeback for a specific order. They tell the agent to report it as fraud, and it updates Sift immediately. ### Banning a repeat offender A trust specialist finds a user who keeps trying to bypass rules. They ask the agent to apply a block_user decision to stop the attempts. ### Auditing a suspicious account A risk analyst wants to see what has happened to a user over the last month. The agent lists the full decision history for review. ## Benefits - Stop fraud faster by using get_user_fraud_score to check risk levels in real time. You no longer have to wait for a manual review or leave your current workspace to see if a transaction is safe. - Keep a clean audit trail of every action taken on an account by using list_user_decision_history for transparency. This makes it easy to see exactly what happened to a user during a dispute. - Take immediate action against bad actors by using apply_user_decision to block users the moment they are flagged. You can move from identifying a threat to neutralizing it in a single step. - Keep your risk data accurate by using report_sift_chargeback to notify Sift of new disputes instantly. This ensures your fraud engine has the most recent data to improve its scoring. - Improve your security model by using track_sift_event to feed the system with more login and transaction data. You can refine your machine learning by logging specific customer behaviors. - Stay on top of your security rules by using list_sift_workflows to see your active fraud prevention setups. This gives you a clear overview of your current security posture at any time. ## How It Works The bottom line is you get a conversational way to manage fraud and chargebacks without leaving your chat app. 1. Subscribe to the Sift (Chargeback) MCP on the Vinkius marketplace. 2. Provide your Sift REST API Key and Account ID in the configuration. 3. Ask your AI agent to check risk scores, report chargebacks, or apply user decisions. ## Frequently Asked Questions **How does the Sift (Chargeback) MCP help with e-commerce fraud?** It gives you a way to check user risk scores and report disputes directly through your AI agent. You can stop bad actors faster by taking action in your chat window instead of navigating a complex dashboard. **Can I block users using the Sift (Chargeback) MCP?** Yes, you can use the Connector to apply manual decisions like blocking a user or accepting an order. Your AI agent handles the request and updates Sift automatically so you can act on risk instantly. **How do I report a new chargeback with Sift (Chargeback)?** Just tell your AI agent to report a chargeback for a specific order. It will send the event to Sift and update the user's risk profile for you, keeping your data current. **Can the Sift (Chargeback) MCP show me a user's history?** It can list the entire history of decisions and labels applied to a specific user. This is great for auditing and seeing how a user's risk profile has changed over time. **Does the Sift (Chargeback) MCP help with risk analysis?** Yes, it lets you pull real-time fraud scores and view your active fraud prevention workflows. You can use this to quickly verify if a transaction is safe or needs a closer look. **Can I block a fraudulent user through the agent?** Yes! Use the `apply_user_decision` tool with the user's ID and the appropriate decision ID (like `block_user`). The action will be applied in Sift immediately. **How do I check the risk score for a specific customer?** Use the `get_user_fraud_score` tool with the customer's unique ID. Your agent will fetch the latest risk score generated by Sift's machine learning models. **Where do I find my Sift API Key and Account ID?** Log in to your Sift Dashboard and navigate to **Settings -> API Keys**. You will find your REST API Key and Account ID there.