Directus MCP Server
# @staminna/directus-mcp-server
MCP server for Directus 12 — items, collections, files, flows, users, and schema tools. TypeScript, typed throughout.
[](https://www.npmjs.com/package/@staminna/directus-mcp-server)
[](https://opensource.org/licenses/MIT)
[](https://github.com/staminna/mcp-server-claude/actions/workflows/ci.yml)
[](https://glama.ai/mcp/servers/staminna/mcp-server-claude)
[](https://glama.ai/mcp/servers/staminna/mcp-server-claude)
### Test Coverage
| Statements | Branches | Functions | Lines |
|------------|----------|-----------|-------|
|  |  |  |  |
Coverage badges are generated from `coverage/coverage-summary.json` by `npm run badges` (no external service required). Run `npm run test:coverage` first.
## Features
- 🔐 **Full Authentication** - Token-based authentication with Directus
- 📦 **Collection Management** - CRUD operations for collections and items
- 📁 **File Operations** - Upload, download, and manage files
- 🔄 **Flow Management** - Create, update, trigger, and manage Directus Flows
- 👥 **User Management** - User CRUD and role management
- 🔍 **Schema Tools** - Analyze and validate collection schemas
- 🩺 **Diagnostics** - Collection access diagnostics and troubleshooting
## Installation
### Via npm (Recommended)
```bash
npm install -g @staminna/directus-mcp-server
```
### From Source
```bash
git clone https://github.com/staminna/mcp-server-claude.git
cd mcp-server-claude
npm install
npm run build
```
## Environment Variables
| Variable | Required | Description |
|----------|----------|-------------|
| `DIRECTUS_URL` | Yes | Your Directus instance URL (e.g., `http://localhost:8065`) |
| `DIRECTUS_TOKEN` | Yes | Static API token with appropriate permissions |
| `DIRECTUS_PROMPTS_COLLECTION_ENABLED` | No | Enable AI prompts collection (`true`/`false`) |
| `DIRECTUS_PROMPTS_COLLECTION` | No | Collection name for AI prompts (default: `ai_prompts`) |
| `DIRECTUS_RESOURCES_ENABLED` | No | Enable resources feature (`true`/`false`) |
| `DIRECTUS_RESOURCES_EXCLUDE_SYSTEM` | No | Exclude system collections from resources (`true`/`false`) |
| `NODE_ENV` | No | Environment mode (`development`/`production`) |
| `DIRECTUS_TIMEOUT` | No | Request timeout in ms (default: `30000`) |
| `DIRECTUS_RETRIES` | No | Retry attempts for network errors, 5xx and 429 (default: `3`) |
| `DIRECTUS_RETRY_DELAY` | No | Base backoff delay in ms (default: `1000`) |
| `DIRECTUS_MAX_RETRY_DELAY` | No | Backoff ceiling in ms (default: `10000`) |
| `DIRECTUS_IMPORT_MAX_FILE_SIZE` | No | Client-side import size ceiling in bytes, mirroring the Directus `IMPORT_MAX_FILE_SIZE` (default: 50 MB) |
| `LOG_LEVEL` | No | `DEBUG`/`INFO`/`WARN`/`ERROR` (default: `INFO`). Logs go to stderr; stdout is reserved for MCP |
### TLS / client certificates
Set these when the Directus instance uses a private CA or requires a client
certificate. Each of `CA`/`CERT`/`KEY`/`PFX` accepts either a file path or the
PEM/DER content itself.
| Variable | Description |
|----------|-------------|
| `DIRECTUS_HTTPS_CA` | Certificate authority |
| `DIRECTUS_HTTPS_CERT` | Client certificate |
| `DIRECTUS_HTTPS_KEY` | Client private key |
| `DIRECTUS_HTTPS_PFX` | PKCS#12 bundle (alternative to cert/key) |
| `DIRECTUS_HTTPS_PASSPHRASE` | Passphrase for the key or PFX |
| `DIRECTUS_HTTPS_REJECT_UNAUTHORIZED` | `false` to accept self-signed certificates |
| `DIRECTUS_HTTPS_SERVERNAME` | SNI server name override |
---
## Authentication — no OAuth required
This server uses a **static Directus access token** (`DIRECTUS_TOKEN`) and runs over **stdio transport**. OAuth is *not* required, by design:
- The MCP specification only defines OAuth 2.1 authorization for **HTTP-based transports**. For stdio servers the spec says implementations *"SHOULD NOT"* use it and should instead retrieve credentials from the environment — exactly what this server does.
- **Directus 12 fully supports static access tokens.** The OAuth 2.1 support Directus added (mid-2026) applies to its own built-in *remote* MCP endpoint and is optional; there are no breaking changes to token authentication in Directus 12 (see `DIRECTUS_V12_BREAKING_CHANGES.md`).
- OAuth only becomes relevant if you expose an MCP server **remotely over HTTP** (Streamable HTTP/SSE). As a local stdio subprocess of Claude Desktop, Claude Code, Cursor, etc., this server needs only the env token.
Generate the token in Directus under **User Settings → Token** (use a dedicated user with least-privilege role for production).
### Using with a Claude subscription (Max/Pro) — no API key needed
MCP servers do not consume Anthropic API tokens themselves; only the AI client's model calls do. If you use this server inside **Claude Code or Claude Desktop with a Claude Max (or Pro) subscription**, the model usage is covered by the subscription — you do **not** need an Anthropic API key. An API key is only required when driving Claude programmatically via the Claude API (e.g. the remote MCP connector).
---
## IDE Configuration
### 🟣 Cursor
1. Open Cursor Settings: `Cmd+,` (macOS) or `Ctrl+,` (Windows/Linux)
2. Search for **"MCP"** or navigate to **Features → MCP Servers**
3. Click **"Edit in settings.json"**
4. Add the following configuration:
```json
{
"mcpServers": {
"directus": {
"command": "npx",
"args": [
"-y",
"@staminna/directus-mcp-server"
],
"env": {
"DIRECTUS_URL": "http://localhost:8065",
"DIRECTUS_TOKEN": "your-directus-token-here"
}
}
}
}
```
**Or if installed locally:**
```json
{
"mcpServers": {
"directus": {
"command": "node",
"args": [
"/path/to/mcp-server-claude/dist/index.js"
],
"env": {
"DIRECTUS_URL": "http://localhost:8065",
"DIRECTUS_TOKEN": "your-directus-token-here"
}
}
}
}
```
5. Save the file and restart Cursor
---
### 🌊 Windsurf
1. Open Windsurf Settings: `Cmd+,` (macOS) or `Ctrl+,` (Windows/Linux)
2. Search for **"MCP Servers"**
3. Click **"Edit in settings.json"**
4. Add the following configuration:
```json
{
"mcpServers": {
"directus": {
"command": "npx",
"args": [
"-y",
"@staminna/directus-mcp-server"
],
"env": {
"DIRECTUS_URL": "http://localhost:8065",
"DIRECTUS_TOKEN": "your-directus-token-here",
"DIRECTUS_PROMPTS_COLLECTION_ENABLED": "true",
"DIRECTUS_PROMPTS_COLLECTION": "ai_prompts",
"DIRECTUS_RESOURCES_ENABLED": "true",
"DIRECTUS_RESOURCES_EXCLUDE_SYSTEM": "true",
"NODE_ENV": "production"
}
}
}
}
```
**Or if installed locally:**
```json
{
"mcpServers": {
"directus": {
"command": "node",
"args": [
"/path/to/mcp-server-claude/dist/index.js"
],
"env": {
"DIRECTUS_URL": "http://localhost:8065",
"DIRECTUS_TOKEN": "your-directus-token-here"
}
}
}
}
```
5. Save the file
6. **Quit Windsurf completely** (`Cmd+Q` or `Ctrl+Q`)
7. Reopen Windsurf and wait ~10 seconds for MCP to initialize
---
### 🤖 Claude Desktop
1. Locate your Claude Desktop config file:
- **macOS**: `~/Library/Application Support/Claude/claude_desktop_config.json`
- **Windows**: `%APPDATA%\Claude\claude_desktop_config.json`
- **Linux**: `~/.config/Claude/claude_desktop_config.json`
2. Create or edit the config file:
```json
{
"mcpServers": {
"directus": {
"command": "npx",
"args": [
"-y",
"@staminna/directus-mcp-server"
],
"env": {
"DIRECTUS_URL": "http://localhost:8065",
"DIRECTUS_TOKEN": "your-directus-token-here"
}
}
}
}
```
**Or if installed locally:**
```json
{
"mcpServers": {
"directus": {
"command": "node",
"args": [
"/path/to/mcp-server-claude/dist/index.js"
],
"env": {
"DIRECTUS_URL": "http://localhost:8065",
"DIRECTUS_TOKEN": "your-directus-token-here"
}
}
}
}
```
3. Save the file and restart Claude Desktop
---
### 🔮 Claude.ai (Web with MCP)
For Claude.ai web interface with MCP support:
1. Navigate to Claude.ai settings
2. Find the MCP configuration section
3. Add a new MCP server with:
```json
{
"name": "directus",
"command": "npx",
"args": ["-y", "@staminna/directus-mcp-server"],
"env": {
"DIRECTUS_URL": "http://localhost:8065",
"DIRECTUS_TOKEN": "your-directus-token-here"
}
}
```
> **Note**: Claude.ai MCP support may require a Pro subscription and specific browser extensions.
---
## Available Tools
### Collection Management
| Tool | Description |
|------|-------------|
| `list_collections` | List all collections in Directus |
| `get_collection_schema` | Get schema for a specific collection |
| `get_collection_items` | Get items from a collection with filtering |
| `create_collection` | Create a new collection |
| `delete_collection` | Delete a collection (requires `confirm`) |
| `create_item` | Create a new item in a collection |
| `update_item` | Update an existing item, optionally into a draft `version` |
| `delete_items` | Delete items by `ids` or by `query` (see note below) |
| `bulk_operations` | Execute bulk create, update, delete |
### Schema & Fields
| Tool | Description |
|------|-------------|
| `create_field` | Create a new field in a collection |
| `update_field` | Update an existing field |
| `delete_field` | Delete a field from a collection |
| `create_relationship` | Create relationships (O2O, O2M, M2O, M2M, M2A) |
| `analyze_collection_schema` | Analyze schema with relationship mapping |
| `validate_collection_schema` | Validate schema and relationships |
| `analyze_relationships` | Analyze relationships across collections |
| `get_schema_snapshot` | Read a full or partial snapshot of the data model |
| `diff_schema` | Compare a snapshot against the live schema (`merge` or `mirror`). Directus drops request bodies over ~96 KB, so pass a partial snapshot from `get_schema_snapshot` with `include_collections` on any sizeable data model — see `DIRECTUS_V12_BREAKING_CHANGES.md` |
| `apply_schema` | Apply a diff (requires `confirm`) |
### Flow Management
| Tool | Description |
|------|-------------|
| `get_flows` | Get all flows with optional filtering |
| `get_flow` | Get a specific flow by ID |
| `create_flow` | Create a new automation flow |
| `update_flow` | Update an existing flow |
| `delete_flow` | Delete a flow |
| `trigger_flow` | Manually trigger a flow |
| `get_operations` | Get flow operations |
### User Management
| Tool | Description |
|------|-------------|
| `get_users` | Get all users with filtering |
| `get_user` | Get a specific user by ID |
### File Management
| Tool | Description |
|------|-------------|
| `get_files` | Get files with filtering and pagination |
| `import_data` | Import CSV/JSON into one collection, or several at once |
### Diagnostics
| Tool | Description |
|------|-------------|
| `diagnose_collection_access` | Diagnose collection access issues |
| `refresh_collection_cache` | Refresh collection cache |
| `validate_collection_creation` | Validate newly created collections |
### Discovery
| Tool | Description |
|------|-------------|
| `search_tools` | Find the tools matching a task description |
### Tool safety annotations
Every tool carries MCP annotations so a client can tell reads from writes before
calling: 17 are `readOnlyHint: true`, 6 are explicitly `destructiveHint: false`
(additive — creates), and 11 are `destructiveHint: true` (deletes, overwriting
updates, `apply_schema`, `import_data`, `trigger_flow`).
Note that `destructiveHint` **defaults to true** in the MCP spec, which is why
the additive tools set it to `false` rather than omitting it.
### Deleting items safely
Following Directus 12.3.0, `delete_items` never falls back to deleting
everything:
- `ids: [...]` deletes those items.
- `query: {...}` deletes everything the query matches.
- Passing both is rejected.
- Passing neither deletes nothing and issues no request.
To delete every item in a collection, ask for it explicitly:
```json
{ "collection": "articles", "query": { "limit": -1 }, "confirm": true }
```
---
## Usage Examples
Once configured, you can interact with Directus through your AI assistant:
```
"List all collections in my Directus instance"
"Create a new collection called 'blog_posts' with title, content, and published fields"
"Get all items from the 'products' collection where status is 'published'"
"Create a new flow that triggers on item creation in the 'orders' collection"
"Analyze the schema of the 'users' collection including relationships"
```
---
## Troubleshooting
### MCP Server Not Connecting
1. **Verify Directus is running**: Ensure your Directus instance is accessible at the configured URL
2. **Check token permissions**: The API token needs appropriate permissions for the operations you want to perform
3. **Restart IDE**: After changing MCP configuration, fully restart your IDE
4. **Check logs**: Look for MCP-related errors in your IDE's developer console
### Permission Errors
Ensure your Directus token has the required permissions:
- Admin token for full access
- Or configure specific role permissions for collections you need to access
### Connection Timeout
If using a remote Directus instance:
- Verify the URL is correct and accessible
- Check firewall/network settings
- Ensure CORS is properly configured on Directus
---
## Development
```bash
# Install dependencies
npm install
# Build
npm run build
# Watch mode
npm run dev
# Run server
npm start
# Type check
npm run typecheck
# Lint
npm run lint
```
### Testing
The project ships unit, integration and end-to-end suites (vitest). Coverage thresholds (95% statements/lines/functions/branches) are enforced — the test run fails below them.
```bash
# Unit + integration tests
npm test
# With coverage report (coverage/ — text, html, lcov, json-summary)
npm run test:coverage
# End-to-end: builds, then spawns the real server over stdio against a mock Directus
npm run test:e2e
# Everything
npm run test:all
# Refresh the README coverage badges from the last coverage run
npm run badges
```
### Live verification against a real Directus
`tests/live/demo.mjs` drives all 34 tools against a real instance over stdio. It is
deliberately outside `npm test` — it needs a credential and a reachable server, so it
is a manual gate rather than a CI one.
```bash
# Read-only + guard phases (touches nothing)
ENV_FILE=.env.mdbaudio npm run test:live
# Also create, mutate and drop a scratch mcp_demo_<stamp> collection
ENV_FILE=.env.mdbaudio npm run test:live -- --write
# Additionally exercise apply_schema, confined to that scratch collection
ENV_FILE=.env.mdbaudio npm run test:live -- --write --apply-schema
```
Credentials are read from `ENV_FILE` (default `.env.mdbaudio`) so they never pass through
shell history. Results are reported per tool as pass / refused-by-instance / fail, keeping
"this server is broken" separate from "this instance declined". `--apply-schema` diffs in
**merge** mode, which yields a strictly additive diff, so it can only re-create the scratch
collection — it cannot drop anything that already existed. Cleanup runs even when an
earlier phase fails.
The e2e suite uses the official MCP SDK client (`StdioClientTransport`) to spawn `dist/index.js` as a subprocess, talking to an in-process mock Directus on an ephemeral port — no real Directus instance or network access needed.
---
## Contributing
Contributions are welcome! Please feel free to submit a Pull Request.
1. Fork the repository
2. Create your feature branch (`git checkout -b feature/amazing-feature`)
3. Commit your changes (`git commit -m 'Add some amazing feature'`)
4. Push to the branch (`git push origin feature/amazing-feature`)
5. Open a Pull Request
---
## License
MIT © [Jorge Domingues Nunes](https://github.com/staminna)
---
## Links
- [npm Package](https://www.npmjs.com/package/@staminna/directus-mcp-server)
- [GitHub Repository](https://github.com/staminna/mcp-server-claude)
- [Directus Documentation](https://docs.directus.io/)
- [MCP Protocol Specification](https://modelcontextprotocol.io/)
TDQS
Scored across 34 tools
Many tools are clearly separated by resource and action, but several clusters overlap: bulk_operations partially duplicates create_item/update_item/delete_items, and analyze_collection_schema/validate_collection_schema/diagnose_collection_access/refresh_collection_cache all deal with schema health and permissions. The descriptions help, but an agent could still misroute requests.
Most tools follow a consistent snake_case verb_noun pattern such as get_collection_items, create_field, update_item, and delete_flow. The few exceptions like bulk_operations and the overlapping schema-related phrases keep it from being perfectly consistent.
At 34 tools, the surface is heavy and includes several narrowly scoped helpers like refresh_collection_cache, validate_collection_creation, and analyze_relationships that could be consolidated. The broad Directus domain justifies some breadth, but this still exceeds the comfortable tool count range.
Core collection, field, item, schema, and flow operations are covered, but there are notable gaps: no single-item getter, no update_collection, no relationship update/delete, no user/file write operations, and flow operations cannot be independently created or updated. Agents can work around some gaps but not all.