Logchef

A Model Context Protocol (MCP) server that connects AI assistants to your Logchef instance.

Logchef is a powerful log management platform that stores logs in ClickHouse, providing fast querying and analysis capabilities. This MCP server enables AI assistants to interact with your Logchef deployment, making log analysis and troubleshooting more accessible through natural conversation.

What You Can Do

With this MCP server, you can ask AI assistants to help you:

• Explore your log infrastructure: See what teams you belong to and what log sources are available
• Query logs effectively: Execute ClickHouse SQL queries to find specific log entries, errors, or patterns
• Understand your data: Get schema information to know what fields are available in your logs
• Analyze log patterns: Generate histograms and time-series data for trend analysis
• Manage saved queries: Create, browse, and update frequently-used queries pinned to a source
• Administer teams and users: Handle team membership, user management, and source configuration (admin users)

Tool Categories

The server provides tools organized into logical categories that can be enabled or disabled as needed:

Profile & Metadata

Core user and server information including team memberships and server version details.

Source Management

Access to log sources across your teams, including schema exploration and source listing.

Log Analysis

The core querying capabilities including SQL execution, histogram generation, and saved query management.

Administration

Team management, user administration, source configuration, and API token management (requires admin privileges).

Use the --disable-<category> flag to turn off tool categories you don't need. For example, --disable-admin removes all administrative tools.

Tools

See docs/tools.md for the full reference including resources, prompts, and annotations.

Tool	Category	Description
get_profile	Profile	Get current user profile
get_teams	Profile	List teams you belong to
get_meta	Profile	Server version and config
get_sources	Sources	All accessible sources across teams
get_team_sources	Sources	Sources for a specific team
query_logs	Logs	Execute ClickHouse SQL (max 100 rows)
get_source_schema	Logs	Column names and types for a source
get_log_histogram	Logs	Time-series histogram with optional grouping
list_saved_queries	Logs	List saved queries (optionally filtered by source_id)
create_saved_query	Logs	Save a new query bound to a source
get_saved_query	Logs	Get a saved query by ID
update_saved_query	Logs	Update an existing saved query (creator-only)
delete_saved_query	Logs	Delete a saved query (creator-only)
query_logchefql	LogchefQL	Execute LogchefQL query (max 500 rows)
translate_logchefql	LogchefQL	Translate LogchefQL to SQL
validate_logchefql	LogchefQL	Validate LogchefQL syntax
get_field_values	Investigate	Top values for a field in a time range
get_log_context	Investigate	Surrounding logs around a timestamp
list_alerts	Investigate	Alert rules for a source
get_alert_history	Investigate	Alert evaluation history
compare_windows	Analysis	Compare query results across two time windows
top_values	Analysis	Top values for multiple fields in one call
generate_query	Discover	Natural language to SQL (requires AI enabled)
get_all_field_dimensions	Discover	Bulk field values for all LowCardinality fields
get_query_telemetry	Telemetry	Query performance from system.query_log
list_all_teams	Admin	List all teams (admin only)
get_team	Admin	Get team details
create_team	Admin	Create a team (admin only)
update_team	Admin	Update team info
delete_team	Admin	Delete a team (admin only)
list_team_members	Admin	List team members
add_team_member	Admin	Add user to team
remove_team_member	Admin	Remove user from team
link_source_to_team	Admin	Grant team access to a source
unlink_source_from_team	Admin	Revoke team access to a source
list_all_users	Admin	List all users (admin only)
get_user	Admin	Get user details (admin only)
create_user	Admin	Create a user (admin only)
update_user	Admin	Update user info (admin only)
delete_user	Admin	Delete a user (admin only)
list_api_tokens	Admin	List your API tokens
create_api_token	Admin	Create an API token
delete_api_token	Admin	Delete an API token
list_all_sources	Admin	List all sources (admin only)
create_source	Admin	Create a source (admin only)
validate_source_connection	Admin	Test ClickHouse connectivity (admin only)
delete_source	Admin	Delete a source (admin only)
get_admin_source_stats	Admin	Source table stats (admin only)

Getting Started

Prerequisites

You'll need:

• A running Logchef instance (learn more at logchef.app)
• A valid Logchef API token with appropriate permissions

Generating an API Token

1. Log into your Logchef instance
2. Navigate to your profile settings
3. Create a new API token with the permissions you need
4. Copy the token for use in the MCP server configuration

Installation

Choose one of the following installation methods:

• Docker image: Use the pre-built Docker image.

  Important: The Docker image's entrypoint is configured to run the MCP server in SSE mode by default, but most users will want to use STDIO mode for direct integration with AI assistants like Claude Desktop:

  1. STDIO Mode: For stdio mode you must explicitly override the default with -t stdio and include the -i flag to keep stdin open:

  docker pull ghcr.io/mr-karan/logchef-mcp:latest
  docker run --rm -i -e LOGCHEF_URL=http://localhost:5173 -e LOGCHEF_API_KEY=<your_api_token> ghcr.io/mr-karan/logchef-mcp:latest -t stdio
  

  2. SSE Mode: In this mode, the server runs as an HTTP server that clients connect to. You must expose port 8000 using the -p flag:

  docker pull ghcr.io/mr-karan/logchef-mcp:latest
  docker run --rm -p 8000:8000 -e LOGCHEF_URL=http://localhost:5173 -e LOGCHEF_API_KEY=<your_api_token> ghcr.io/mr-karan/logchef-mcp:latest
  

  3. Streamable HTTP Mode: In this mode, the server operates as an independent process that can handle multiple client connections. You must expose port 8000 using the -p flag:

  docker pull ghcr.io/mr-karan/logchef-mcp:latest
  docker run --rm -p 8000:8000 -e LOGCHEF_URL=http://localhost:5173 -e LOGCHEF_API_KEY=<your_api_token> ghcr.io/mr-karan/logchef-mcp:latest -t streamable-http
  

• Download binary: Download the latest release of logchef-mcp from the releases page and place it in your $PATH.

• Build from source: If you have a Go toolchain installed you can also build and install it from source:

  just build
  

  Or manually:

  go build -o logchef-mcp.bin ./cmd/logchef-mcp
  

3. Add the server configuration to your client configuration file. For example, for Claude Desktop:

   If using the binary:

   {
     "mcpServers": {
       "logchef": {
         "command": "logchef-mcp.bin",
         "args": [],
         "env": {
           "LOGCHEF_URL": "http://localhost:5173",
           "LOGCHEF_API_KEY": "<your_api_token>"
         }
       }
     }
   }
   

Note: if you see Error: spawn logchef-mcp.bin ENOENT in Claude Desktop, you need to specify the full path to logchef-mcp.bin.

If using Docker:

{
  "mcpServers": {
    "logchef": {
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "-i",
        "-e",
        "LOGCHEF_URL",
        "-e",
        "LOGCHEF_API_KEY",
        "ghcr.io/mr-karan/logchef-mcp:latest",
        "-t",
        "stdio"
      ],
      "env": {
        "LOGCHEF_URL": "http://localhost:5173",
        "LOGCHEF_API_KEY": "<your_api_token>"
      }
    }
  }
}

Note: The -t stdio argument is essential here because it overrides the default SSE mode in the Docker image.

Using VSCode with remote MCP server

If you're using VSCode and running the MCP server remotely, you can configure it to connect via HTTP or SSE transport. Make sure your .vscode/settings.json includes the following:

For HTTP transport (recommended):

"mcp": {
  "servers": {
    "logchef": {
      "type": "http",
      "url": "http://localhost:8000/mcp",
      "headers": {
        "X-Logchef-URL": "https://your-logchef-instance.com",
        "X-Logchef-API-Key": "your_api_token_here"
      }
    }
  }
}

For SSE transport:

"mcp": {
  "servers": {
    "logchef": {
      "type": "sse",
      "url": "http://localhost:8000/sse"
    }
  }
}

Using HTTP Headers for Authentication

When running the MCP server in SSE or streamable-http mode, you can pass Logchef credentials via HTTP headers instead of environment variables. This is useful when the server needs to handle multiple clients with different Logchef instances or API keys.

The server recognizes these headers:

• X-Logchef-URL: The Logchef instance URL
• X-Logchef-API-Key: The API token for authentication

Example configuration for clients that support custom headers:

{
  "mcpServers": {
    "logchef": {
      "type": "http",
      "url": "http://localhost:8000/mcp",
      "headers": {
        "X-Logchef-URL": "https://your-logchef-instance.com",
        "X-Logchef-API-Key": "your_api_token_here"
      }
    }
  }
}

If headers are not provided, the server will fall back to environment variables (LOGCHEF_URL and LOGCHEF_API_KEY).

Debug Mode

You can enable debug mode for the Logchef transport by adding the -debug flag to the command. This will provide detailed logging of HTTP requests and responses between the MCP server and the Logchef API, which can be helpful for troubleshooting.

To use debug mode with the Claude Desktop configuration, update your config as follows:

If using the binary:

{
  "mcpServers": {
    "logchef": {
      "command": "logchef-mcp.bin",
      "args": ["-debug"],
      "env": {
        "LOGCHEF_URL": "http://localhost:5173",
        "LOGCHEF_API_KEY": "<your_api_token>"
      }
    }
  }
}

If using Docker:

{
  "mcpServers": {
    "logchef": {
      "command": "docker",
      "args": [
        "run",
        "--rm",
        "-i",
        "-e",
        "LOGCHEF_URL",
        "-e",
        "LOGCHEF_API_KEY",
        "ghcr.io/mr-karan/logchef-mcp:latest",
        "-t",
        "stdio",
        "-debug"
      ],
      "env": {
        "LOGCHEF_URL": "http://localhost:5173",
        "LOGCHEF_API_KEY": "<your_api_token>"
      }
    }
  }
}

Note: As with the standard configuration, the -t stdio argument is required to override the default SSE mode in the Docker image.

Tool Configuration

You can selectively enable or disable tool categories using command-line flags:

• --enabled-tools: Comma-separated list of tool categories to enable (default: all)
• --disable-profile: Disable profile management tools
• --disable-sources: Disable source management tools
• --disable-logs: Disable log querying tools
• --disable-logchefql: Disable LogchefQL tools
• --disable-investigate: Disable investigation tools
• --disable-admin: Disable admin tools
• --disable-analysis: Disable analysis tools (compare_windows, top_values)
• --disable-telemetry: Disable telemetry tools
• --disable-discover: Disable discovery tools (AI query generation, field dimensions)

Example with selective tool enabling:

{
  "mcpServers": {
    "logchef": {
      "command": "logchef-mcp.bin",
      "args": ["--enabled-tools", "profile,logs"],
      "env": {
        "LOGCHEF_URL": "http://localhost:5173",
        "LOGCHEF_API_KEY": "<your_api_token>"
      }
    }
  }
}

Working with Logchef Through AI

This MCP server enables you to interact with your Logchef logs through natural conversation with AI assistants. Here's how it typically works:

Discovery Workflow

1. "What log sources do I have access to?" → The AI uses get_sources to show your available sources
2. "What data is in the nginx source?" → The AI calls get_source_schema to explain the log structure
3. "Show me recent errors" → The AI constructs and executes a ClickHouse query using query_logs

Practical Examples

• Troubleshooting: "Find all 500 errors in the last hour from the web service logs"
• Analysis: "Show me a histogram of log volume over the past day"
• Investigation: "What are the most common error messages in the database logs?"
• Monitoring: "Create a saved query for tracking API response times"

AI-Assisted Query Building

The AI assistant understands ClickHouse SQL and can help you:

• Build complex queries with proper syntax
• Optimize queries for better performance
• Explain what fields are available in your log data
• Suggest useful queries based on common log analysis patterns

Since Logchef uses ClickHouse as the storage backend, you get the full power of ClickHouse's analytical capabilities through natural language interaction.

Development

This project is written in Go and uses Just as a command runner for common development tasks.

Prerequisites

• Go 1.21 or later
• A running Logchef instance for testing
• Valid Logchef API credentials

Quick Start

# Install dependencies
just deps

# Run in STDIO mode (default for development)
just run

# Run in SSE mode for web-based clients
just run-sse

# Build the binary
just build

# Run tests
just test

Environment Variables

For development, set these environment variables:

export LOGCHEF_URL=http://localhost:5173  # Your Logchef instance URL
export LOGCHEF_API_KEY=your_api_token_here

Docker Development

Build and test the Docker image locally:

# Build development image
just build-image

# Test GoReleaser Docker setup
just build-goreleaser-image

# Run in different modes
just docker-run-stdio
just docker-run-sse

Testing & Verification

Test the server with your Logchef instance:

# Build and run with your credentials
just build
LOGCHEF_URL=http://localhost:5173 LOGCHEF_API_KEY=your_token ./dist/logchef-mcp

# Check version info
just show-version

# Test basic functionality
just help

Contributing

Contributions are welcome! When adding new tools or features:

1. Follow the existing patterns for tool definition and client methods
2. Add appropriate error handling and validation
3. Update the tools table in this README
4. Test with a real Logchef instance

Please open an issue to discuss larger changes before implementing them.

License

This project is licensed under the Apache License, Version 2.0.

The design of this repository was inspired by mcp-grafana.