# Tracelane MCP

> Tracelane MCP is a freemium MCP server for developer tools. Read-only MCP access to Tracelane agent traces and guardrail verdicts for debugging. It runs locally on your machine and is started by your AI client.

- **Kind:** MCP server
- **Pricing:** Freemium
- **Starting price:** $29 a month
- **Platforms:** CLI, self-hosted, API
- **Website:** https://tracelane.dev/docs/mcp-server
- **Repository:** https://github.com/tracelane/tracelane/tree/HEAD/apps/mcp
- **MCP Registry name:** io.github.tracelane/tracelane-mcp
- **Categories:** developer tools, AI models and LLM tooling
- **Listing updated:** 2026-11-16

## About

Tracelane is an observability and guardrail platform for LLM agents, and this server lets Claude Desktop, Claude Code or Cursor inspect its data while you debug. It is read-only by design: no write tools exist, every read is scoped to your tenant, and tool arguments cannot override the tenant or inject raw SQL.

There are two modes. By default it reads through Tracelane's gateway with just an API key; setting CLICKHOUSE_URL switches to self-host mode and reads your own ClickHouse. Its explain_guardrail_block tool turns a flag into a readable account, such as a lethal-trifecta sequence where untrusted input, private data access and an outbound write converge across spans. Trace replay and eval results are also exposed.

It runs over stdio from npm as @tracelanedev/mcp, or serves Streamable HTTP when you run it yourself; no hosted MCP endpoint exists.

## Features

- **Trace browsing:** List recent traces and filter by model or errors.
- **Span detail:** Fetch a single span with its LLM attributes.
- **Free-text search:** Search span names and attributes across traces.
- **Guardrail explanations:** Explains why a request was flagged or blocked.
- **Eval results:** Lists evals and returns a given result.
- **Trace replay:** Replays a trace, optionally including tool calls.
- **Tenant isolation:** Reads are tenant-scoped and non-2xx replies surface as errors.

## Good for

- Debug why an agent got stuck in a loop
- Review which span triggered a guardrail flag
- Find all traces with errors for one model
- Check an eval result from your editor

## Install

### Claude Code (Terminal)

```sh
claude mcp add tracelane-mcp --env TRACELANE_API_KEY='<TRACELANE_API_KEY>' --env TRACELANE_GATEWAY_URL='<TRACELANE_GATEWAY_URL>' --env CLICKHOUSE_PASSWORD='<CLICKHOUSE_PASSWORD>' -- npx -y @tracelanedev/mcp
```

### Claude Desktop (claude_desktop_config.json)

```json
{
  "mcpServers": {
    "tracelane-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@tracelanedev/mcp"
      ],
      "env": {
        "TRACELANE_API_KEY": "<TRACELANE_API_KEY>",
        "TRACELANE_GATEWAY_URL": "<TRACELANE_GATEWAY_URL>",
        "CLICKHOUSE_PASSWORD": "<CLICKHOUSE_PASSWORD>"
      }
    }
  }
}
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "tracelane-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@tracelanedev/mcp"
      ],
      "env": {
        "TRACELANE_API_KEY": "<TRACELANE_API_KEY>",
        "TRACELANE_GATEWAY_URL": "<TRACELANE_GATEWAY_URL>",
        "CLICKHOUSE_PASSWORD": "<CLICKHOUSE_PASSWORD>"
      }
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "tracelane-mcp": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@tracelanedev/mcp"
      ],
      "env": {
        "TRACELANE_API_KEY": "<TRACELANE_API_KEY>",
        "TRACELANE_GATEWAY_URL": "<TRACELANE_GATEWAY_URL>",
        "CLICKHOUSE_PASSWORD": "<CLICKHOUSE_PASSWORD>"
      }
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "tracelane-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@tracelanedev/mcp"
      ],
      "env": {
        "TRACELANE_API_KEY": "<TRACELANE_API_KEY>",
        "TRACELANE_GATEWAY_URL": "<TRACELANE_GATEWAY_URL>",
        "CLICKHOUSE_PASSWORD": "<CLICKHOUSE_PASSWORD>"
      }
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "tracelane-mcp": {
      "command": "npx",
      "args": [
        "-y",
        "@tracelanedev/mcp"
      ],
      "env": {
        "TRACELANE_API_KEY": "<TRACELANE_API_KEY>",
        "TRACELANE_GATEWAY_URL": "<TRACELANE_GATEWAY_URL>",
        "CLICKHOUSE_PASSWORD": "<CLICKHOUSE_PASSWORD>"
      }
    }
  }
}
```

## Tools

- `list_traces`: Lists recent traces with optional model and error filters.
- `get_trace`: Returns all spans for a trace.
- `get_span`: Returns full details for a span.
- `search_traces`: Free-text search across span names and attributes.
- `explain_guardrail_block`: Explains a guardrail signal or blocked request.
- `list_evals`: Lists available evals.
- `get_eval_result`: Returns the result of an eval.
- `replay_trace`: Replays a trace, optionally with tool calls.

## FAQ

### Is it read-only?

Yes. No write tools are exposed.

### Do I need ClickHouse?

No. Gateway mode needs only TRACELANE_API_KEY; ClickHouse is for self-hosting.

### Is there a hosted MCP URL?

No. The README says mcp.tracelane.dev does not resolve, so you run it yourself.

### What does Tracelane cost?

The core is Apache 2.0 and free to self-host; hosted tiers start at $29 a month (Builder), with a free hosted tier for non-commercial use.

### Is Tracelane MCP free?

Tracelane MCP has a free plan, with paid plans for more usage or features. Paid plans start at $29 a month.

### How do I add Tracelane MCP to Claude Code?

Run this command in your terminal: claude mcp add tracelane-mcp --env TRACELANE_API_KEY='<TRACELANE_API_KEY>' --env TRACELANE_GATEWAY_URL='<TRACELANE_GATEWAY_URL>' --env CLICKHOUSE_PASSWORD='<CLICKHOUSE_PASSWORD>' -- npx -y @tracelanedev/mcp

### What are the best alternatives to Tracelane MCP?

The closest alternatives to Tracelane MCP are REA, Codebase Memory, GitHub, Serena MCP: the IDE for your agent and XERJ.

## Alternatives

- [REA](https://ai-powered.com/mcp/rea): REA is an open-source MCP server for developer tools.
- [Codebase Memory](https://ai-powered.com/mcp/codebase-memory): Codebase Memory is an open-source MCP server for knowledge and memory.
- [GitHub](https://ai-powered.com/mcp/github-2): GitHub is an open-source MCP server for developer tools.
- [Serena MCP: the IDE for your agent](https://ai-powered.com/mcp/serena-mcp-the-ide-for-your-agent): Serena MCP: the IDE for your agent is an open-source MCP server for developer tools.
- [XERJ](https://ai-powered.com/mcp/xerj): XERJ is an open-source MCP server for knowledge and memory.
- [Ouroboros](https://ai-powered.com/mcp/ouroboros): Ouroboros is an open-source MCP server for developer tools.
- [Claude Code Ultimate Guide](https://ai-powered.com/mcp/claude-code-ultimate-guide): Claude Code Ultimate Guide is an open-source MCP server for ai models and llm tooling.
- [Concord MCP](https://ai-powered.com/mcp/concord-mcp): Concord MCP is a freemium MCP server for developer tools.

---
Source: https://ai-powered.com/mcp/tracelane-mcp
