@dugleelabs/strapi-mcp-server

A general-purpose, open-source MCP server for Strapi v4. Exposes Strapi CRUD and schema introspection as MCP tools, with optional AI content generation (research + draft) via the Vercel AI SDK and configurable web search providers. Use it with Claude Desktop, Cursor, or any MCP-compatible agent to manage content through natural language — or automate content creation pipelines entirely.

Prerequisites

• Node.js 20+
• pnpm (or npm/npx)
• A running Strapi v4 instance with an API token

Quick Start

Copy the env vars and run with pnpm dlx:

STRAPI_URL=http://localhost:1337 \
STRAPI_API_TOKEN=your-token-here \
pnpm dlx @dugleelabs/strapi-mcp-server

The server starts over stdio and is ready to use immediately. All configuration is via environment variables — no install step required.

Claude Desktop Configuration

Add this to your claude_desktop_config.json (usually at ~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "strapi": {
      "command": "pnpm",
      "args": ["dlx", "@dugleelabs/strapi-mcp-server"],
      "env": {
        "STRAPI_URL": "http://localhost:1337",
        "STRAPI_API_TOKEN": "your-token-here",
        "AI_PROVIDER": "anthropic",
        "AI_MODEL": "claude-sonnet-4-6",
        "ANTHROPIC_API_KEY": "sk-ant-...",
        "SEARCH_PROVIDER": "tavily",
        "TAVILY_API_KEY": "tvly-..."
      }
    }
  }
}

Environment Variables

| Variable | Required | Description | Example | |---|---|---|---| | STRAPI_URL | Yes | Base URL of your Strapi instance | http://localhost:1337 | | STRAPI_API_TOKEN | Yes | Strapi API token (Full Access or scoped) | abc123... | | SEARCH_PROVIDER | No | Enables research_topic: tavily, brave, or exa | tavily | | TAVILY_API_KEY | If SEARCH_PROVIDER=tavily | Tavily API key | tvly-... | | BRAVE_API_KEY | If SEARCH_PROVIDER=brave | Brave Search API key | BSA... | | EXA_API_KEY | If SEARCH_PROVIDER=exa | Exa API key | exa-... | | AI_PROVIDER | No | Enables AI tools: openai, anthropic, google, mistral, ollama | anthropic | | AI_MODEL | If AI_PROVIDER set | Model name for the selected provider | claude-sonnet-4-6 | | ANTHROPIC_API_KEY | If AI_PROVIDER=anthropic | Anthropic API key | sk-ant-... | | OPENAI_API_KEY | If AI_PROVIDER=openai | OpenAI API key | sk-... | | GOOGLE_GENERATIVE_AI_API_KEY | If AI_PROVIDER=google | Google AI API key | AIza... | | MISTRAL_API_KEY | If AI_PROVIDER=mistral | Mistral API key | ... | | OLLAMA_BASE_URL | If AI_PROVIDER=ollama | Ollama base URL | http://localhost:11434 |

Tools

Tier 1 — CRUD (always enabled)

list_entries

List entries from a Strapi content type.

{
  "contentType": "articles",
  "page": 1,
  "pageSize": 25,
  "status": "draft",
  "filters": { "title": { "$contains": "AI" } }
}

Returns: { entries, total, page, pageCount }

get_entry

Fetch a single entry by ID.

{ "contentType": "articles", "id": 42 }

Returns: { entry }

create_entry

Create a new entry. Use get_content_type_schema first to see available fields.

{
  "contentType": "articles",
  "data": { "title": "My Post", "content": "..." },
  "publish": false
}

Returns: { id, adminUrl, entry }

update_entry

Partially update an existing entry.

{
  "contentType": "articles",
  "id": 42,
  "data": { "title": "Updated Title" }
}

Returns: { id, entry }

delete_entry

Permanently delete an entry.

{ "contentType": "articles", "id": 42 }

Returns: { success: true, message: "Entry 42 deleted from articles" }

list_content_types

List all content types in the Strapi instance. Use this to discover available types before CRUD operations.

No input required.

Returns: { contentTypes: [{ uid, displayName, pluralName, kind }] }

get_content_type_schema

Get the field schema for a content type — field names, types, and required status.

{ "uid": "api::article.article" }

Returns: { uid, displayName, attributes: { [field]: { type, required, default } } }

Tier 2 — Search (requires SEARCH_PROVIDER)

research_topic

Search the web for current information on a topic.

{
  "topic": "TypeScript 2026 trends",
  "context": "developer tooling",
  "maxResults": 10
}

Returns: { query, results: [{ title, url, content, score }], provider }

Tier 3 — AI Content (requires AI_PROVIDER)

generate_draft

Generate a structured blog post draft using the configured AI provider.

{
  "topic": "TypeScript 2026 trends",
  "researchResults": [...],
  "styleGuide": "Write tersely for senior engineers.",
  "targetWordCount": 800
}

Returns: { title, body, metaDescription, tags, wordCount }

create_content_from_research

End-to-end: research a topic → generate a draft → save to Strapi. Requires both search and AI configured.

{
  "topic": "TypeScript 2026 trends",
  "contentType": "articles",
  "context": "focus on tooling and DX",
  "styleGuide": "Terse, developer-focused.",
  "fieldMapping": {
    "title": "title",
    "body": "content",
    "metaDescription": "seo_description",
    "tags": "tags"
  }
}

Note on fieldMapping: Every Strapi instance has different field names. Provide fieldMapping to match your content type's actual fields. If omitted, draft keys are used as-is (title, body, metaDescription, tags).

Returns: { strapiId, adminUrl, title, wordCount, step }

MCP Prompts

Three prompts are registered as slash commands in MCP clients:

| Prompt | Arguments | Description | |---|---|---| | dugleelabs/create-blog-from-topic | topic (required), contentType (optional) | Research a topic and create a draft blog post in Strapi | | dugleelabs/list-drafts | contentType (required) | Show all unpublished drafts for a content type | | dugleelabs/review-entry | contentType, id (both required) | Fetch an entry and suggest improvements before publishing |

Search Providers

Tavily

SEARCH_PROVIDER=tavily TAVILY_API_KEY=tvly-...

Get an API key at tavily.com.

Brave Search

SEARCH_PROVIDER=brave BRAVE_API_KEY=BSA...

Get an API key at brave.com/search/api.

Exa

SEARCH_PROVIDER=exa EXA_API_KEY=exa-...

Get an API key at exa.ai.

AI Providers

Install only the provider package you need:

Anthropic (Claude)

pnpm add @ai-sdk/anthropic
AI_PROVIDER=anthropic AI_MODEL=claude-sonnet-4-6 ANTHROPIC_API_KEY=sk-ant-...

OpenAI

pnpm add @ai-sdk/openai
AI_PROVIDER=openai AI_MODEL=gpt-4o OPENAI_API_KEY=sk-...

Google

pnpm add @ai-sdk/google
AI_PROVIDER=google AI_MODEL=gemini-1.5-pro GOOGLE_GENERATIVE_AI_API_KEY=AIza...

Mistral

pnpm add @ai-sdk/mistral
AI_PROVIDER=mistral AI_MODEL=mistral-large-latest MISTRAL_API_KEY=...

Ollama (local)

pnpm add @ai-sdk/openai
AI_PROVIDER=ollama AI_MODEL=llama3 OLLAMA_BASE_URL=http://localhost:11434

Known Limitations

• Schema introspection requires Full Access token: The list_content_types and get_content_type_schema tools use Strapi's content-type-builder API, which may return 403 with restricted tokens. CRUD tools work with any scoped token that has the appropriate permissions.

• pnpm dlx latency: Running via pnpm dlx downloads the package on every invocation, which adds ~2–5 seconds of startup latency in Claude Desktop. To avoid this, install globally: pnpm add -g @dugleelabs/strapi-mcp-server.

• Strapi v4 only: Targets the Strapi v4 REST API. Strapi v5 support is planned.

• No content type assumptions: The server does not assume any specific field names. Always use get_content_type_schema or provide fieldMapping to match your Strapi schema.

Contributing

Issues and PRs welcome at github.com/dugleelabs/strapi-mcp-server.

git clone [email protected]:dugleelabs/strapi-mcp-server.git
cd strapi-mcp-server
pnpm install
cp .env.example .env  # fill in STRAPI_URL and STRAPI_API_TOKEN
pnpm test
pnpm build

Use MCP Inspector for interactive local testing:

pnpm dlx @modelcontextprotocol/inspector pnpm start

License

MIT — see LICENSE.