EventWhisper

EventWhisper

Enables fast, scriptable access to Windows .evtx event logs for incident response and digital forensics. Supports filtering events by time windows, Event IDs, and keywords with field projection to reduce output size.

Category
访问服务器

README

EventWhisper — A Windows Event Log MCP

Tests Coverage Python License: GPLv3 Platform

EventWhisper exposes fast, scriptable access to Windows .evtx logs via an MCP (Model Context Protocol) server. It’s pure Python (uses the evtx library), instead of a PowerShell wrapper and does not execute commands on the host — safer for incident response, digital forensics, and threat hunting.

The server exposes two primary tools:

  • List EVTX files (optionally recursive) for any directory.
  • Filter events to directly search for specific records.

Built specifically for IR/DFIR & hunting so you don’t need to look up Event IDs all day. Designed, developed, and tested on Windows with Claude — and usable from any MCP client.


Features

  • Targeted filtering: time window, EventID(s), and case-insensitive keywords (include/exclude)
  • Field projection: return only the dotted paths you need (huge output reduction)
  • Input normalization: accepts flexible input formats so it still works if types/format aren’t perfect.
  • MCP-ready: plugs into Claude Desktop (and other MCP clients)

Demo

https://github.com/user-attachments/assets/e1878c86-56f6-466b-bbc4-04d1edeeb36b


Requirements (Windows)

EventWhisper can be set up with Poetry, whether you run it inside Claude or standalone. On Windows, make sure Poetry is on your PATH.

Install Poetry (PowerShell):

(Invoke-WebRequest -Uri https://install.python-poetry.org -UseBasicParsing).Content | py -

Add Poetry to PATH (user install):

C:\Users\<YourUser>\AppData\Roaming\Python\Scripts

Verify:

poetry --version

Install & Run

Primarily developed for use with Claude, but you can run the MCP server standalone and plug it into other LLMs.

Clone and install:

git clone https://github.com/hexastrike/eventwhisper
cd eventwhisper
poetry install

Run the MCP server:

poetry run python -m eventwhisper.mcp.server
# (or) plain Python if deps are installed globally:
python -m eventwhisper.mcp.server

Claude MCP configuration

Add this to Claude Desktop’s MCP config (e.g., %AppData%\Claude\claude_desktop_config.json):

{
  "mcpServers": {
    "EventWhisper": {
      "type": "stdio",
      "command": "poetry",
      "args": [
        "-C",
        "C:\\Path\\To\\eventwhisper",
        "run",
        "python",
        "-m",
        "eventwhisper.mcp.server"
      ],
      "env": { "PYTHONIOENCODING": "utf-8" }
    }
  }
}

Open Settings → Developer to confirm EventWhisper is registered and “running”.

Claude MCP Config

When prompted, allow the tool:

Claude Allow Tool

Try a first prompt:

Use EventWhisper to list .evtx files in C:\Windows\System32\winevt\Logs.


Development

EventWhisper was written to be simple and maintainable. We use pytest for tests and ruff for linting/formatting. Run everything locally (or via pre-commit):

poetry install
poetry run ruff format .
poetry run ruff check . --fix
poetry run pytest --cov=eventwhisper --cov-report=term-missing
# pre-commit on all files:
poetry run pre-commit run --all-files
# pre-push hooks (full tests + coverage):
poetry run pre-commit run --hook-stage push --all-files

When making pull requests, ensure all tests pass and coverage thresholds are met. Every new utility or feature should include clear tests and follow PEP 8. Tests live in the tests/ directory.

Core layout & API

Primary logic lives in eventwhisper/evtxio/evtxio.py — EVTX iteration/filtering/projection.

The core server function is get_events_from_evtx() (a wrapper around the iterator iter_events_from_evtx()):

def get_events_from_evtx(
    provider: str | Path,
    start: datetime | str | None = None,
    end: datetime | str | None = None,
    results_limit: int | str | None = RESULTS_LIMIT,
    event_ids: int | str | Sequence[int] | Sequence[str] | None = None,
    contains: str | Sequence[str] | None = None,
    not_contains: str | Sequence[str] | None = None,
    fields: str | Sequence[str] | None = None,
) -> list[str]:

Constants live in eventwhisper/utils/config.py and include both:

  • RESULTS_LIMIT — max number of events returned (LLMs have token/char caps).
  • SCAN_LIMIT — max events scanned to keep responses fast on huge logs.

Example Python usage

# example.py
from __future__ import annotations
from datetime import datetime, timezone
from eventwhisper.evtxio.evtxio import get_events_from_evtx

# 1) First matching event from Security log
print(get_events_from_evtx(r"C:\Windows\System32\winevt\Logs\Security.evtx", results_limit=1))

# 2) Filter by keywords and project fields
events = get_events_from_evtx(
    r"C:\Windows\System32\winevt\Logs\Security.evtx",
    contains=["powershell", "Invoke-"],
    fields=["Event.System.EventID", "Event.EventData.Image", "Event.EventData.CommandLine"],
    results_limit=5,
)
for e in events:
    print(e)

# 3) Time-bounded query (UTC)
start = datetime(2025, 1, 1, tzinfo=timezone.utc)
end = datetime(2025, 1, 31, 23, 59, 59, tzinfo=timezone.utc)
print(len(get_events_from_evtx(r"C:\Windows\System32\winevt\Logs\Security.evtx", start=start, end=end)))

Run it with poetry run python example.py.

Troubleshooting

  • Big logs: results are capped by RESULTS_LIMIT and scanning is bounded to keep responses fast. If your MCP client doesn’t realize results are capped, run a second, narrower query.
  • “Blocked” files: if you downloaded .evtx from the Internet, right-click → PropertiesUnblock.
  • Paths: in Python, prefer raw strings for Windows paths, e.g., r"C:\path\file.evtx".
  • README video: GitHub doesn’t render <video>; use a thumbnail → MP4 link (see Demo section).

Roadmap

  • Faster scanning for very large .evtx files
  • Better normalization for malformed/corrupted EVTX
  • Convenience utilities (e.g., quick overview: counts, unique EventIDs, time range)
  • Optional paging / multi-query strategies for huge datasets

License

Distributed under the GPLv3. See LICENSE.

推荐服务器

Baidu Map

Baidu Map

百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。

官方
精选
JavaScript
Playwright MCP Server

Playwright MCP Server

一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。

官方
精选
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。

官方
精选
本地
TypeScript
VeyraX

VeyraX

一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。

官方
精选
本地
graphlit-mcp-server

graphlit-mcp-server

模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。

官方
精选
TypeScript
Kagi MCP Server

Kagi MCP Server

一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

Exa MCP Server

模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。

官方
精选