Drupal MCP
Enter a valid instance URL (https:// is required unless it is localhost).
cinatra install cinatra-ai/drupal-mcp-connector
Drupal MCP
Let Cinatra agents read, draft, and publish content on your Drupal sites. Connect one or more Drupal instances and your agents can browse recent nodes, create draft revisions, update fields, and push updates live.
Install: add the connector from the Cinatra marketplace. Your Drupal site must run drupal/mcp_tools ^1.0 with its Streamable HTTP endpoint (/_mcp_tools) reachable from the Cinatra host. Create a Nango connection for the cinatra-drupal integration (provider: private-api-bearer), then open Settings → Integrations → Drupal to register each instance by site URL and Nango connection ID.
Usage: agents discover instances via drupal_instances_list; ask an agent to list recent nodes, create a draft, update fields, or publish. drupal_content_editor_run turns a plain-language instruction into the full draft-revision workflow.
Configuration: each instance needs a siteUrl and a nangoConnectionId; no plaintext credentials are stored — the Bearer token lives only in the Nango vault, resolved at call time.
Architecture: the connector owns the Drupal MCP client and instance-settings store and registers the drupal-mcp and widget-auth capabilities itself at activation — the Cinatra core ships no Drupal client code. In dev, its cinatra.devSetup hook provisions the local Drupal fixture on boot.
API notes: drupal_node_get proxies via the recent-content list (100 most-recent nodes). drupal_node_update needs a draft revision — call drupal_node_create_draft_revision first. Empty-string fields are stripped to prevent accidental wipes.
Development: run pnpm vitest run --no-coverage. See AGENTS.md for tool-name mapping and invariants.
Troubleshooting: if drupal_status reports an instance unreachable, verify /_mcp_tools is reachable and the Nango connection is active; a 401 means the Bearer token expired — regenerate it and update the Nango connection.
Works with
- Drupal 10 and 11 with the
drupal/mcp_toolsmodule installed
Capabilities
- Browse and read recent Drupal nodes from inside an agent flow
- Draft a new node of any configured content type
- Update fields on an existing node through a clean draft revision
- Publish a draft to make it live
- Edit a Drupal node from a plain-language instruction
- Chat with an in-CMS widget that edits the open node in the Drupal editor
- Receive a webhook notification when a node is published on a connected site
More from Cinatra
The built-in fallback type that guarantees every item in the library has at least one valid classification. When a file, connector reference, or dashboard…
Enter a valid instance URL (https:// is required unless it is localhost).
cinatra install cinatra-ai/default-artifact
The conversational skill bundle that powers the Cinatra chat assistant. It shapes how the assistant talks, recognizes what you are trying to do, and…
Enter a valid instance URL (https:// is required unless it is localhost).
cinatra install cinatra-ai/assistant-skills
A quick checkpoint that runs right before any drafting step. It shows you which skills will guide the work — tone, brand voice, writing…
Enter a valid instance URL (https:// is required unless it is localhost).
cinatra install cinatra-ai/skill-recommender-agent
Connect your Anthropic account so Cinatra agents can run on Claude. Once you bring your key, any agent pinned to a Claude model becomes…
Enter a valid instance URL (https:// is required unless it is localhost).
cinatra install cinatra-ai/anthropic-connector
Connect OpenAI so every agent in your workspace can run on GPT models. Holds your API key, default model, and service tier behind a…
Enter a valid instance URL (https:// is required unless it is localhost).
cinatra install cinatra-ai/openai-connector
Link a YouTube account to your workspace so agents can read channel uploads through the YouTube Data API. The connection reuses your existing Google…
Enter a valid instance URL (https:// is required unless it is localhost).
cinatra install cinatra-ai/youtube-connector
Reviews
There are no reviews yet.