Gives Claude direct access to the MITRE ATT&CK framework for querying adversarial tactics and techniques. You get six tools: query_technique and search_technique_full for looking up attack methods by ID or fuzzy name search, query_mitigations and query_detections for defensive measures, list_tactics for the full taxonomy, and server_info for version details. It runs either as stdio for local integration or HTTP for remote deployment. Useful when you're analyzing threat intelligence, writing security documentation, or need to quickly reference technique IDs like T1059.001 without leaving your workflow. The data comes from the enterprise-attack.json dataset and includes subtechniques, kill chain phases, and platform applicability.
This project is an MCP (Model Context Protocol) server for querying ATT&CK (Adversarial Tactics, Techniques, and Common Knowledge) techniques and tactics. It provides a way to access and retrieve information about various attack techniques and tactics used by adversaries.
The server provides the following tools:
technique_id (string, optional): The ID of the technique to query.tech_name (string, optional): The name (or partial name) of the technique to query. 支持名称模糊搜索。{
"technique_id": "T1059.001"
}
{
"tech_name": "phishing"
}
{ "results": [...], "count": N } 的字典,其中 results 为匹配技术完整数据列表。
technique_id (string, optional): 要查询的技术ID。tech_name (string, optional): 技术名称关键字,支持模糊匹配。{
"technique_id": "T1059.001"
}
{
"tech_name": "phishing"
}
technique_id (string, required): 要查询的技术ID{
"technique_id": "T1059.001"
}
technique_id (string, required): 要查询的技术ID{
"technique_id": "T1059.001"
}
{}
To use this MCP server, you need to have an MCP client configured to connect to it. Once connected, you can use the provided tools to query ATT&CK techniques and tactics.
python main.py
ATTACK_MCP_MODE=stdio),适用于 Smithery、Cursor 等支持本地 MCP stdio 的客户端。使用 CLI 参数切换模式:
python main.py --mode http --host 0.0.0.0 --port 8081 --log-level info
或通过环境变量控制:
export ATTACK_MCP_MODE=http
export ATTACK_MCP_HOST=0.0.0.0 # 可选,默认 0.0.0.0 或 $HOST
export ATTACK_MCP_PORT=8081 # 可选,默认 8081 或 $PORT
export ATTACK_MCP_LOG_LEVEL=info # 可选,默认 info
python main.py
运行后服务以 streamable HTTP 方式暴露,可在客户端配置服务类型为 "http",地址如 http://127.0.0.1:8081/mcp。
远程部署(如 Smithery Cloud)通常会提供 PORT 或 MCP_TRANSPORT 环境变量,可直接运行 python main.py 即使用 HTTP。对于值为 streaming、streamable、streamable-http、streamable HTTP transport 或 stdioNotSupported 等新枚举的运行环境,程序会自动回退到 HTTP 模式,无需额外配置。
Smithery 等容器平台会通过 PORT(默认为 8081)告知监听端口;程序会自动读取该值并监听在 0.0.0.0:$PORT。
工具名称:query_technique、search_technique_full、query_mitigations、query_detections、list_tactics、server_info
参数示例:
{
"technique_id": "T1059.001"
}
{
"tech_name": "phishing"
}
search_technique_full 获取技术的完整详细信息:
{
"tech_name": "phishing"
}
{
"technique_id": "T1059.001"
}
{
"technique_id": "T1059.001"
}
{}
{}
具体的客户端配置方式请参考您的 MCP 客户端文档,将上述服务地址和工具名称填入对应位置即可。
pip install -r requirements.txt.ATT&CK is a curated knowledge base and model for cyber adversary behavior, reflecting the various phases of an adversary's attack lifecycle and the platforms they are known to target. ATT&CK is useful for understanding security risks against any specific technology or organization.
pip install -r requirements.txt
python main.py
python main.py --mode http --host 127.0.0.1 --port 8081
docker build -t attack-mcp-server .
docker run -p 8081:8081 attack-mcp-server
如有问题请联系维护者。
io.github.ericm1018/skillfm-llm-cost-optimizer-openai-anthropic-usage
io.github.mikerawsonnz/llm-orchestration-agent
io.github.mikerawsonnz/authenticated-llm-agent
labforgedev/copilot-memory-mcp
csoai-org/agent-prompt-injection-firewall-mcp
io.github.mikerawsonnz/authenticated-multi-llm-agent