mcp-solid-pod
MCP server for reading, searching, and optionally writing data in Solid Pods via LDP/RDF with headless client-credentials authentication.
README
mcp-solid-pod
MCP server for Solid Pods — list, read, search, and (opt-in) write — with CSS client-credentials auth for headless use (AIsuru MCP Gateway, Claude Desktop, etc.).
Portions adapted from @jeswr/solid-mcp (MIT).
Requirements
- Node.js >= 24
- A Solid server that supports client credentials (e.g. Community Solid Server)
Tools
| Tool | Args | Notes |
|---|---|---|
solid_list |
{ container } |
List LDP container children |
solid_read |
{ url } |
Read resource (Turtle for RDF) |
solid_search |
{ query, scope? } |
Client-side search |
solid_write |
{ url, content, contentType } |
PUT — disabled unless SOLID_MCP_READONLY=false |
All URLs are confined to SOLID_MCP_POD_ROOT (SSRF / pod-scope guard).
Environment
| Variable | Required | Description |
|---|---|---|
SOLID_MCP_POD_ROOT |
yes | Pod/container URL ending in / |
SOLID_MCP_OIDC_ISSUER |
for private pods | IdP base (e.g. http://localhost:3000/) |
SOLID_MCP_CLIENT_ID |
for private pods | CSS client credentials id |
SOLID_MCP_CLIENT_SECRET |
for private pods | CSS client credentials secret |
SOLID_MCP_WEBID |
no | Enables Type-Index search hints |
SOLID_MCP_READONLY |
no | Default true |
Credentials are read from the environment only (gateway server_parameters) — never from the LLM.
Local run
npm install
npm run build
npm test
SOLID_MCP_POD_ROOT=http://localhost:3000/alice/ \
SOLID_MCP_OIDC_ISSUER=http://localhost:3000/ \
SOLID_MCP_CLIENT_ID=… \
SOLID_MCP_CLIENT_SECRET=… \
node build/cli.js
E2E against Community Solid Server
npm run build
npm run e2e
This starts npx @solid/community-server with e2e/seed.json, creates client credentials, and asserts list/read/search/write via the MCP stdio client.
With Docker instead:
docker compose -f e2e/docker-compose.yml up -d
SKIP_CSS_START=1 npm run e2e
docker compose -f e2e/docker-compose.yml down
Configure in the AIsuru MCP Gateway UI (admin)
Gateway admin → MCP Server Schemas → Nuovo schema. Fill the form with the values below (copy-paste).
Informazioni base
| Field | Value |
|---|---|
| ID Schema | solid-pod-mcp-server |
| Nome | Solid Pod |
| Tipo Server | Server MCP Generico (generic) |
| Logo/Emoji | 🌐 |
Comando/URL di Esempio (url_command_hidden) |
npx -y @memori.ai/mcp-solid-pod |
| Priorità | 16 |
| Schema disponibile | checked |
Local unpublished package (dev gateway host):
node /absolute/path/to/MCP Solid/build/cli.js
Descrizioni
🇮🇹 Descrizione Italiana
Server MCP per leggere e (opzionalmente) scrivere dati in un Solid Pod tramite LDP/RDF. Autenticazione headless con client credentials CSS / Solid-OIDC. Le credenziali non sono mai passate al LLM.
🇺🇸 English Description
MCP Server to read and (optionally) write data in a Solid Pod via LDP/RDF. Headless auth with CSS / Solid-OIDC client credentials. Credentials are not given to the LLM.
Prompt per Agenti AI
Prompt per Lista Tools
Questo server Solid Pod permette di elencare, leggere e cercare risorse LDP/RDF nel pod configurato. Questa funzione permette di leggere la lista di tool disponibili per questo MCP.
Prompt per Esecuzione Tools
Usa solid_list per esplorare un container (URL che termina con /), solid_read per leggere una risorsa, solid_search per cercare per nome/URL/letterali RDF. solid_write e' disponibile solo se SOLID_MCP_READONLY=false. Tutti gli URL devono stare dentro SOLID_MCP_POD_ROOT.
Parametri (JSON) — paste into “Parametri di Configurazione”
[
{
"name": "SOLID_MCP_POD_ROOT",
"type": "string",
"required": true,
"description": "URL assoluto del pod/container (deve terminare con /)",
"description_it": "URL assoluto del pod/container (deve terminare con /)",
"description_en": "Absolute pod/container URL (must end with /)",
"example": "https://pod.example/alice/"
},
{
"name": "SOLID_MCP_OIDC_ISSUER",
"type": "string",
"required": true,
"description": "Base URL dell'Identity Provider / CSS (es. https://pod.example/)",
"description_it": "Base URL dell'Identity Provider / CSS (es. https://pod.example/)",
"description_en": "Identity Provider / CSS base URL (e.g. https://pod.example/)",
"example": "https://pod.example/"
},
{
"name": "SOLID_MCP_CLIENT_ID",
"type": "string",
"required": true,
"description": "Client credentials ID (CSS)",
"description_it": "Client credentials ID (CSS)",
"description_en": "Client credentials ID (CSS)",
"example": "my-token_xxxxxxxx"
},
{
"name": "SOLID_MCP_CLIENT_SECRET",
"type": "string",
"required": true,
"description": "Client credentials secret (CSS)",
"description_it": "Client credentials secret (CSS)",
"description_en": "Client credentials secret (CSS)",
"example": "your-client-secret"
},
{
"name": "SOLID_MCP_WEBID",
"type": "string",
"required": false,
"description": "WebID del proprietario (abilita Type-Index in search)",
"description_it": "WebID del proprietario (abilita Type-Index in search)",
"description_en": "Owner WebID (enables Type-Index hints in search)",
"example": "https://pod.example/alice/profile/card#me"
},
{
"name": "SOLID_MCP_READONLY",
"type": "boolean",
"required": false,
"description": "Se true (default) disabilita solid_write",
"description_it": "Se true (default) disabilita solid_write",
"description_en": "If true (default) disables solid_write",
"example": "true",
"default": "true"
}
]
Click Crea Schema.
When an agent activates this MCP
Fill instance parameters with real values (injected as env vars; never sent to the LLM):
{
"SOLID_MCP_POD_ROOT": "https://pod.example/alice/",
"SOLID_MCP_OIDC_ISSUER": "https://pod.example/",
"SOLID_MCP_CLIENT_ID": "my-token_xxxxxxxx",
"SOLID_MCP_CLIENT_SECRET": "the-secret-value",
"SOLID_MCP_WEBID": "https://pod.example/alice/profile/card#me",
"SOLID_MCP_READONLY": "true"
}
SOLID_MCP_CLIENT_SECRET is encrypted at rest.
License
MIT — see LICENSE.
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。