truenas-mcp
Read-only MCP server for TrueNAS Scale debugging, providing k3s tools for kubectl operations and whitelisted midclt calls for system info, apps, and pools.
README
TrueNAS MCP Server
Read-only MCP server for TrueNAS Scale debugging. Runs directly on TrueNAS and provides:
- k3s tools — kubectl get/describe/logs/events/top wrapped as MCP tools
- TrueNAS API — whitelisted
midclt callmethods for system info, apps, pools
No destructive operations. Every tool is read-only.
Quick Start
# Install
pip install git+https://github.com/a-simple-test-organization/truenas-mcp.git
# Run (no auth)
MCP_PORT=38888 python -m truenas_mcp.server
# Starts on http://0.0.0.0:38888/mcp
# Health check
curl http://localhost:38888/health
# → {"status": "ok", "auth_enabled": false}
Authentication
Set MCP_TOKEN to enable Bearer token auth. Then all requests must include Authorization: Bearer <token>.
# Generate a token
openssl rand -hex 32
# Run with auth
MCP_TOKEN="your-token-here" MCP_PORT=38888 python -m truenas_mcp.server
Client configuration (e.g. OpenClaw, Claude Code, OpenCode):
{
"mcpServers": {
"truenas": {
"url": "http://192.168.2.48:38888/mcp",
"headers": {
"Authorization": "Bearer your-token-here"
}
}
}
}
The /health endpoint is always available without authentication.
systemd Deployment
# Copy the unit file
cp deploy/truenas-mcp.service /etc/systemd/system/
# Edit Environment=MCP_TOKEN= to set your token
vim /etc/systemd/system/truenas-mcp.service
# Enable and start
systemctl daemon-reload
systemctl enable --now truenas-mcp.service
# Check
systemctl status truenas-mcp
journalctl -u truenas-mcp -f
MCP Tools
k3s (9 tools)
| Tool | Description |
|---|---|
kubectl_get |
Get resources (pods, deployments, nodes, ...) |
kubectl_describe |
Detailed resource description |
kubectl_logs |
Container logs |
kubectl_events |
Cluster events |
kubectl_nodes |
Node status |
kubectl_top_pods |
Resource usage by pod |
kubectl_top_nodes |
Resource usage by node |
kubectl_api_resources |
Available API types |
pod_status_summary |
Quick all-pods overview |
TrueNAS (2 tools)
| Tool | Description |
|---|---|
midclt_call |
Call whitelisted TrueNAS API method |
midclt_call_arg |
Call with arguments |
Debugging Slow App Startup
1. pod_status_summary
2. kubectl_events(all_namespaces=True)
3. kubectl_describe(resource_type="pod", name="<slow-pod>", namespace="ix-<app>")
4. kubectl_logs(pod_name="<slow-pod>", namespace="ix-<app>")
5. midclt_call(method="chart.release.query")
Environment Variables
| Variable | Default | Description |
|---|---|---|
K3S_BIN |
auto (PATH + fallbacks) | Path to k3s binary |
MIDCLT_BIN |
auto (PATH + fallbacks) | Path to midclt binary |
MCP_PORT |
8000 |
HTTP listen port |
MCP_HOST |
0.0.0.0 |
HTTP listen host |
MCP_PATH |
/mcp |
Streamable HTTP endpoint path |
MCP_TOKEN |
(empty = no auth) | Bearer token for authentication |
License
MIT
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。
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。
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。
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。
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。
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。
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。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。