RELATED 01
Mindwtr MCP Server
OFFICIAL SERVERRead and edit tasks and projects in the Mindwtr GTD app from an MCP client
2.4kGitHub stars
723/wknpm downloads
54popularity score
Hygraph MCP Server
Hosted server for reading and editing content and schema in Hygraph headless CMS projects
Official · hygraph.com Remote (hosted) · Streamable HTTP Developer toolsProductivity
Connect your client
Remote (hosted)
Add in one click
Opens the app, which asks before it adds the server.
Or add it by command or config
Claude CodeTerminal
claude mcp add --transport http hygraph https://mcp.hygraph.com/mcpClaude (desktop and web)Settings → Connectors → Add custom connector
https://mcp.hygraph.com/mcpChatGPTSettings → Apps → Advanced → Developer mode, then Create app
https://mcp.hygraph.com/mcpCursor~/.cursor/mcp.json
{
"mcpServers": {
"hygraph": {
"url": "https://mcp.hygraph.com/mcp"
}
}
}VS Code.vscode/mcp.json
{
"servers": {
"hygraph": {
"type": "http",
"url": "https://mcp.hygraph.com/mcp"
}
}
}Codex~/.codex/config.toml
[mcp_servers.hygraph]
url = "https://mcp.hygraph.com/mcp"Gemini CLITerminal
gemini mcp add --scope user --transport http hygraph https://mcp.hygraph.com/mcpWhat it does
Hygraph runs this MCP server itself, so there is nothing to install: a client connects to one of two hosted addresses. The global endpoint at https://mcp.hygraph.com/mcp signs in with your Hygraph login and sees every project your user can open; the assistant first calls list_projects and list_environments, then names the project and environment on each later call. The project endpoint, https://mcp-{REGION}.hygraph.com/{PROJECT_ID}/{ENVIRONMENT}/mcp, is fixed to one project and environment and takes a Permanent Auth Token as a bearer header, so its reach is whatever that token allows.
Through it an assistant can browse content types and their field shapes, list, fetch, sample and full-text search entries, run Content API GraphQL queries and mutations, create and update draft entries and publish them. On the schema side it can read a project overview and apply batched schema migrations in a single transaction, with a dry run option. It can also load the project's AI guidelines (brand voice, glossary, audience, legal) before writing, and start manual Hygraph agents on entries and check their progress.
The tools reference lists 19 tools in all: 17 on the project endpoint and two discovery tools that exist only on the global endpoint. The agent tools are conditional: trigger_agents and check_progress appear only when the environment has at least one active manual agent (Hygraph MCP tools reference).
What it needs
A Hygraph account with access to the project. For the global endpoint, a client that supports OAuth sign-in; on Claude Team and Enterprise plans an Owner may first have to add the Hygraph connector in organization settings. For the project endpoint, a Permanent Auth Token (generated under Project Settings > Access) and the project's MCP endpoint URL; clients that only speak stdio (Cursor, Windsurf, Claude Desktop) bridge to it with mcp-remote, which needs Node.js 20.18.1 or later.
Limits
Deleting entries or schema elements and unpublishing entries are not offered; execute_graphql rejects delete* and unpublish* operations, so those steps stay manual in Hygraph Studio. Agents bound to workflows cannot be started from MCP, and agent output lands in DRAFT rather than being published. AI guidelines are read-only. search_content is case-insensitive with no operators. The docs note that Cursor's default Auto model mode does not work with this server; a model must be chosen manually.
Who it suits
Teams already running content on Hygraph who want an assistant to inspect models, draft and publish entries, run bulk updates or apply schema migrations without writing scripts. The project endpoint with a read-only or content-only token suits shared or production setups where reach should be narrow; the global endpoint suits one person moving between several projects. It is of no use without a Hygraph project, and anyone who needs code that runs on their own machine should look elsewhere, since only the vendor-hosted service is offered.
Tools 19
list_projects- List the projects the signed-in user can open (global endpoint only)
list_environments- List environment names in a project (global endpoint only)
list_entity_types- List the content type names in the environment
get_entity_schema- Show fields, types and required flags for one content type
list_entities- List entries of one type, with limit, stage and field choice
get_entities_by_id- Fetch given entries by ID, across types in one call
discover_entities- Sample a few entries from several types for exploration
search_content- Full-text search across all content types
execute_graphql- Run a Content API query or mutation, except delete and unpublish
create_entry- Create a new entry as a draft
update_entry- Change an existing entry in draft
publish_entry- Publish an entry from draft
get_project_info- Overview of models, components, enums, locales, stages and migration operations
get_management_operation_schema- Show the input shape and an example for one migration operation
submit_batch_migration- Apply schema changes in one transaction, with an optional dry run
get_ai_guidelines- Load the project's brand, glossary, audience and legal guidelines
list_agents- List agents in the environment and which can be run manually
trigger_agents- Start a manual agent on one or more entries without waiting
check_progress- Report the latest agent run status per entry