"""
mcpstat - Usage tracking and analytics for MCP servers.
https://github.com/tekkidev/mcpstat
Copyright (c) 2026 Vadim Bakhrenkov
SPDX-License-Identifier: MIT
A pure-stdlib Python package for adding SQLite-backed usage statistics,
optional file logging, and built-in analytics to Model Context Protocol servers.
## Quick Start
```python
from mcp.server import Server
from mcpstat import MCPStat
app = Server("my-server")
stat = MCPStat("my-server") # That's it!
@app.call_tool()
async def handle_tool(name: str, arguments: dict):
await stat.record(name, "tool")
# ... your logic
```
## Features
- SQLite-backed usage tracking for tools, prompts, resources
- Optional file-based audit logging
- Built-in stats query functions
- Stats prompt generator for LLM consumption
- Metadata enrichment (tags, descriptions)
- Async-first, thread-safe design
- No required dependencies (pure Python stdlib)
## Configuration
All settings are passed at initialization:
```python
stat = MCPStat(
server_name="my-server",
db_path="./my_mcp_data.sqlite",
log_path="./my_mcp.log",
log_enabled=True,
metadata_presets={
"my_tool": {"tags": ["api"], "short": "My tool description"}
}
)
```
Environment variables (optional override):
- `MCPSTAT_DB_PATH`: SQLite database path
- `MCPSTAT_LOG_PATH`: Log file path
- `MCPSTAT_LOG_ENABLED`: Enable file logging (true/1/yes)
"""
from __future__ import annotations
__version__ = "0.2.1"
__author__ = "Vadim Bakhrenkov"
__license__ = "MIT"
from mcpstat.core import MCPStat
from mcpstat.database import MCPStatDatabase
from mcpstat.logging import MCPStatLogger
from mcpstat.prompts import build_prompt_definition, generate_stats_prompt
from mcpstat.tools import BuiltinToolsHandler, build_tool_definitions
from mcpstat.utils import derive_short_description, normalize_tags
__all__ = [
"BuiltinToolsHandler",
"MCPStat",
"MCPStatDatabase",
"MCPStatLogger",
"__version__",
"build_prompt_definition",
"build_tool_definitions",
"derive_short_description",
"generate_stats_prompt",
"normalize_tags",
]
