MCP User Guide
Connect an MCP-compatible AI client to read and work with Packet Capture Pro data after you approve the connection.
01What Is MCP?
MCP (Model Context Protocol) is a standard protocol that connects AI clients to local applications.
When the Packet Capture Pro MCP service is enabled, MCP-compatible AI clients can read and work with data in Packet Capture Pro after you approve the connection. With MCP, you can use natural-language instructions to ask an AI client to:
- View and filter live capture data.
- View and filter capture history.
- Create, edit, enable, or delete rewrite rules.
- Run standard and advanced replays.
- Create and run debugging sessions.
- Import and analyze HAR files.
02Requirements
Before using MCP with Packet Capture Pro, make sure that:
- Packet Capture Pro has been updated to a version that supports MCP (version 3.0.0 or later).
- Your AI client supports a local MCP server that uses stdio transport.
- The MCP service is enabled in Packet Capture Pro.
- The Packet Capture Pro MCP server has been added to your AI client.
- You approved the AI client when it connected to Packet Capture Pro for the first time.
03Opening the MCP Service
04MCP Service Switch
4.1 Turning On the MCP Service
When you turn on MCP Service, Packet Capture Pro begins accepting MCP connection requests from AI clients.
Turning on the service does not immediately send data to an AI client. You must also configure the AI client and approve it the first time it connects. After a connection is established, the AI client can discover and call the MCP tools provided by Packet Capture Pro.
4.2 Turning Off the MCP Service
When you turn off the MCP service:
- All connected AI clients are disconnected.
- AI clients can no longer read data from Packet Capture Pro.
- AI clients can no longer call the MCP tools provided by Packet Capture Pro.
- MCP calls that are waiting for a result may be terminated or return a connection error.
05Connection Status
The MCP service page shows the current connection status. The Connected Clients section lists the AI clients that are currently connected.
| Status | Description |
|---|---|
| Not connected | The MCP service is on, but no AI client is currently connected. |
| Connecting | An AI client is establishing a connection to Packet Capture Pro. |
| Connected | The AI client is connected and can call the tools provided by Packet Capture Pro. |
06First-Time Connection Approval
When a new AI client requests access to Packet Capture Pro for the first time, you must choose whether to allow the connection. After you approve the connection, the client can call all MCP tools provided by Packet Capture Pro.
The AI client may have its own tool-confirmation, security-approval, or automatic-execution settings. These settings are controlled by the AI client and are not managed by Packet Capture Pro.
07Sensitive Data Protection
Sensitive Data Protection reduces the risk of authentication information in captured traffic being sent directly to an AI client.
When Sensitive Data Protection is enabled, Packet Capture Pro identifies and masks the following sensitive information before returning data to an AI client. Detection is based on field names, data location, and content patterns:
| Data type | Default handling |
|---|---|
Authorization | Keeps the field name and masks the field value. |
Cookie / Set-Cookie | Keeps cookie names and masks their values. |
Token | Masks values in commonly used headers when a token is detected. |
Sensitive Data Protection affects only the data returned to AI clients through MCP. It does not modify the original capture data stored in Packet Capture Pro. When Sensitive Data Protection is disabled, Packet Capture Pro no longer masks these sensitive fields automatically before returning data to an AI client.
08Configuring an AI Client
After you select an AI client, Packet Capture Pro displays the relevant setup instructions and the path to the MCP server executable.
Follow the on-screen instructions to add the Packet Capture Pro MCP server to your AI client.
09MCP Tool Reference
After the connection is established, the AI client can discover and call the following Packet Capture Pro tools.
| Category | Tool | Description |
|---|---|---|
| Live capture | pcp_capture_live_get_status | Get the current live-capture status. |
| Live capture | pcp_capture_live_set_enabled | Start or stop live capture. |
| Live capture | pcp_capture_live_filter | Filter live-capture records by specified criteria. |
| Live capture | pcp_capture_live_get_by_id | Get live-capture details by ID. |
| Live capture | pcp_capture_live_generate_curl | Generate a cURL command from a record. |
| Live capture | pcp_capture_live_generate_har | Generate a HAR file from a record. |
| Live capture | pcp_capture_live_delete | Delete a live-capture record by ID. |
| Live capture | pcp_capture_live_clear | Clear all live-capture records. |
| Live capture | pcp_capture_live_wait_for_match | Wait for a new capture record that matches specified criteria. |
| Capture history | pcp_capture_history_filter | Filter capture-history records by specified criteria. |
| Capture history | pcp_capture_history_get_by_id | Get a capture-history record by ID. |
| Capture history | pcp_capture_history_generate_curl | Generate a cURL command from a capture-history record. |
| Capture history | pcp_capture_history_generate_har | Generate a HAR file from a capture-history record. |
| Capture history | pcp_capture_history_delete | Delete a specified capture-history record. |
| Capture history | pcp_capture_history_clear | Clear all capture-history records. |
| Rewrite | pcp_rewrite_list | Get the list of rewrite rules. |
| Rewrite | pcp_rewrite_get_by_id | Get a rewrite rule by ID. |
| Rewrite | pcp_rewrite_create | Create a rewrite rule from scratch or from captured data. |
| Rewrite | pcp_rewrite_update | Update a specified rewrite rule. |
| Rewrite | pcp_rewrite_set_enabled | Enable or disable a specified rewrite rule. |
| Rewrite | pcp_rewrite_delete | Delete a specified rewrite rule. |
| Replay | pcp_replay_execute | Replay a request using its original content. |
| Replay | pcp_replay_get_result | Get the replay result. |
| Advanced replay | pcp_advanced_replay_execute | Configure advanced replay options and run the replay for live or historical capture data. |
| Advanced replay | pcp_advanced_replay_get_result | Get the advanced replay result. |
| Debug | pcp_debug_create | Create a debugging session from scratch or from captured data. |
| Debug | pcp_debug_update | Update a debugging session. |
| Debug | pcp_debug_execute | Run a debugging session. |
| Debug | pcp_debug_get_result | Get debugging results and execution details. |
| Debug | pcp_debug_delete | Delete a specified debugging session. |
| Import | pcp_import_file | Import a HAR file. |
| Import | pcp_import_get_result | Get the import result. |
| Import | pcp_import_get_by_id | Get import details by ID. |
10Tool Execution Rules
Packet Capture Pro follows a “tool availability defines access” model.
- Any tool listed in the tool reference can be called by an approved, connected AI client.
- Individual tools cannot be enabled or disabled separately.
- Packet Capture Pro does not display a separate confirmation prompt for every tool call.
- Before executing a tool, Packet Capture Pro validates required parameters, target objects, file-access scope, and concurrency limits.
- If parameters are invalid, the target does not exist, or execution fails, Packet Capture Pro returns a clear error to the AI client.
- The AI client may still require your approval through its own tool-approval workflow.
- Results of actions such as deleting or clearing data, replaying requests, and applying rewrites are not automatically rolled back if a rule is later disabled or the MCP connection is closed.
11Important Notes
- MCP tools can perform real actions, including capturing traffic, rewriting and replaying requests, deleting data, debugging, and importing files.
- Only allow AI clients you trust to connect to Packet Capture Pro.
- Data that has been deleted or cleared may not be recoverable through MCP.
- A replay request cannot be canceled after it has been sent.
- Requests already processed by a rewrite rule are not restored when that rule is disabled or deleted.
- Turning off the MCP service disconnects all connected AI clients.
- The AI client may require confirmation before calling a tool.
- MCP setup, tool-confirmation behavior, and timeout limits vary by AI client.
- Multi-step tasks also depend on the model, context window, and execution limits.
RealSignal Pte. Ltd. © 2026 All rights reserved