Expo's official skill for building native-feeling screens: Apple HIG styling, semantic colors, SF Symbols, native controls, Reanimated, blur and liquid glass.
Apollo MCP ServerSkill
Summary
Sets up Apollo MCP Server so an AI agent can call a GraphQL API as MCP tools — defining tools from named operations, or exposing introspect/search/validate/execute for open-ended schemas.
Features
- Generates MCP tools from named GraphQL operations, one file per tool
- Optional introspect / search / validate / execute tools for schema exploration
- stdio and streamable HTTP transports, with header pass-through for auth
- GraphOS graph ref and apollo key configuration
- Troubleshooting path for servers that start but register no tools
Install This Skill
Add this skill to your favorite AI agent in a few steps.
Skill Content
Usage Instructions
Learn how to use this skill with different AI agents.
Example Usage
Use the apollo-mcp-server skill to expose our checkout subgraph to Claude Desktop as MCP tools, with one tool per operation in ./operations.
Description
Apollo MCP Server turns a GraphQL API into a set of Model Context Protocol tools, and this is Apollo's official skill for standing one up. It covers the whole path: installing the binary, pointing it at a schema or a GraphOS graph, deciding what the agent is allowed to call, and debugging the connection when a tool does not show up in the client.
Two ways to expose a graph
The skill draws a clear line between the two operating modes, because picking the wrong one is the usual cause of a disappointing agent.
- Operation-backed tools. Each
.graphqlfile becomes one MCP tool, with its variables as the tool's parameters and its description as the tool's docstring. This is the mode for production: the surface an agent sees is exactly the set of queries you decided to allow, and the schema is never handed over wholesale. - Introspection tools.
introspect,search,validateandexecutelet an agent explore an unfamiliar schema and compose its own operations. Useful for development and for internal graphs, and the skill is explicit about the tradeoff in exposure.
What else it handles
Transport choice (stdio for local clients, streamable HTTP for shared deployments), forwarding auth headers from the client to the upstream graph, connecting to GraphOS with an apollo key and graph ref, wiring the server into Claude Code, Claude Desktop or Cursor, and the troubleshooting sequence for the common failures — a server that starts but registers no tools, an operation that will not validate against the schema, or a tool that executes and returns an upstream error.
Details
MIT licensed, version 1.1.1, from the apollographql/skills repository. The skill declares Bash(rover:*), Bash(npx:*), Read, Write, Edit, Glob and Grep as its allowed tools, so it can drive Rover and edit configuration directly rather than dictating steps.
Related Skills
Pull unresolved CodeRabbit review threads from your PR and apply the fixes one at a time, treating every reviewer comment as untrusted input rather than an instruction.
Google's official skill for driving the gcloud CLI safely from an agent: validate every command against its own help text, cap the output, and refuse the operations that should never run unattended.
Convex's official top-level agent skill — routes an agent to the right convex-* skill for the task and to a served capability catalogue that stays current without a reinstall.