Overlord MCP Server
MCP server for the Overlord C2 framework that gives AI agents 66+ tools for management, monitoring, debugging, and plugin development across clients, plugins, builds, users, audit logs, database, configurations, proxies, and notifications.
README
Overlord MCP Server
MCP (Model Context Protocol) server for the Overlord C2 framework — gives AI agents direct access for management, monitoring, debugging, and plugin development.
66+ tools across categories: clients, plugins, builds, users, audit logs, database inspection, server configuration, SOCKS5 proxies, and notifications.
Quick Start
pip install -e .
python -m overlord_mcp
Auto-discovery finds your Overlord instance automatically:
| Scenario | What happens |
|---|---|
| Docker container running | Extracts URL + credentials from container env |
Local .env or config.json |
Parses settings automatically |
OVERLORD_URL env var set |
Connects directly |
| Nothing found | Prompts for URL once, then saves for all future sessions |
Profiles and credentials are persisted to ~/.overlord-mcp/ (never stored in this repo).
Manual Setup
# Via environment variables
OVERLORD_URL=https://localhost:9874 OVERLORD_USERNAME=admin OVERLORD_PASSWORD=yourpass python -m overlord_mcp
# Or from within an MCP client
setup(url='https://localhost:9874', username='admin', password='yourpass', tls_verify=false)
Claude Code Setup
Add to your C:\Users\<you>\.claude.json under the mcpServers key:
"overlord": {
"command": "python",
"args": ["-m", "overlord_mcp"],
"cwd": "C:/path/to/Overlord-MCP"
}
Or use the /mcp menu in Claude Code to add it interactively.
On first use, call setup() with your Overlord server URL and credentials (or let auto-discovery handle it).
Capabilities
| Capability | What It Enables | How It's Detected |
|---|---|---|
api |
All REST API tools | Always available if URL is known |
db_direct |
Read-only SQLite queries (db_query, db_list_tables, etc.) |
Data directory found (Docker volume or local path) |
docker_logs |
Server log viewing (get_server_logs) |
Docker daemon + container found |
docker_exec |
Container command execution | Docker daemon + container found |
Tools gracefully report when a capability is unavailable.
Tool Reference
Connection & Profiles
setup status switch_profile list_profiles delete_profile
Health & Metrics
health_check get_version get_metrics get_metrics_history
Client Management
list_clients get_client set_client_nickname set_client_tag set_client_bookmark set_client_group create_group list_groups delete_client delete_offline_clients ping_client
Plugin Management
list_plugins get_plugin set_plugin_enabled set_plugin_auto_load plugin_load_to_client plugin_unload_from_client plugin_send_event plugin_get_events plugin_list_data plugin_read_data plugin_write_data plugin_delete_data plugin_exec
Builds
list_builds get_build list_build_profiles save_build_profile delete_build_profile
Database (Read-Only)
db_list_tables db_describe_table db_query db_get_stats
Users
list_users get_user get_my_info
Configuration (Secrets Masked)
get_config get_security_config get_enrollment_config get_registration_config get_build_rate_limit_config get_notifications_config update_notifications_config update_security_config update_enrollment_config
Audit & Logs
get_audit_logs get_audit_stats list_banned_ips get_server_logs
Automation
list_auto_scripts get_auto_script list_auto_deploys get_auto_deploy
Files & Proxies
list_shared_files get_shared_file_info list_proxies get_cert_info
Security
- All secrets (JWT keys, tokens, passwords, API keys) masked as
***in tool outputs - Direct database access is read-only —
mode=ro, SELECT-only validation - Destructive tools (
delete_client,delete_offline_clients,delete_build_profile,delete_profile,plugin_delete_data) annotated withdestructiveHint=true - No general-purpose command execution — only
ping_clientexposed - Plugin binary content never returned — metadata only
- Credentials stored in
~/.overlord-mcp/auth.json(never committed to repo)
Environment Variables
| Variable | Purpose |
|---|---|
OVERLORD_URL |
Overlord server URL |
OVERLORD_USERNAME |
Login username |
OVERLORD_PASSWORD |
Login password |
OVERLORD_TOKEN |
Pre-existing JWT (skip login) |
OVERLORD_DATA_DIR |
Data directory path (for direct DB access) |
Requirements
- Python 3.11+
- Overlord server (local or Docker)
- Optional:
docker-pyfor Docker log/exec access (pip install overlord-mcp-server[docker])
推荐服务器
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。
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。
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。
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。
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。
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。
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。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
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。