Mirrors MCP server
{
"mcpServers": {
"runmirrors-mirrors": {
"type": "http",
"url": "https://api.runmirrors.com/mcp"
}
}
}{
"mcpServers": {
"runmirrors-mirrors": {
"type": "http",
"url": "https://api.runmirrors.com/mcp"
}
}
}{
"mcpServers": {
"runmirrors-mirrors": {
"type": "http",
"url": "https://api.runmirrors.com/mcp"
}
}
}{
"servers": {
"runmirrors-mirrors": {
"type": "http",
"url": "https://api.runmirrors.com/mcp"
}
}
}{
"mcpServers": {
"runmirrors-mirrors": {
"type": "http",
"url": "https://api.runmirrors.com/mcp"
}
}
}file .mcp.json (in your project root)file claude_desktop_config.jsonfile ~/.cursor/mcp.jsonfile .vscode/mcp.jsonfile ~/.codeium/windsurf/mcp_config.jsonversion 1.0.0
What it does
A Go library that collects LLM agent traces and sends them to the Mirrors backend for replay testing. It is for developers building LLM agents in Go who want to test their agents against real production sessions in a staging environment.
- Record traces from your LLM agent and send them to Mirrors for regression testing against rebuilt staging environments.
- Capture multi-turn conversations with tool calls and tool results to replay and verify agent behavior.
- Batch and flush traces in the background without blocking your application.
Tools this server exposes
| Tool | Does |
|---|---|
| LogTrace | Record a sequence of messages (system, user, assistant, tool calls, tool results) as a single trace and queue it for sending. |
| Init | Initialize the trace collector with an API key, project name, and optional configuration for batching and endpoint settings. |
| Flush | Block until all queued traces are sent to the Mirrors backend. |
| Shutdown | Stop the background goroutine that sends traces, with a timeout to ensure in-flight traces are flushed first. |
Before you enable it
An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.