Learning Commons MCP Server

Learning Commons Knowledge Graph is a hosted MCP server that gives AI agents structured access to K-12 academic standards, the granular learning components that compose them, and the progressions that connect them. It is built on top of the Learning Commons Knowledge Graph, which combines the Common Core State Standards plus standards for 18 individual U.S. states (sourced from 1EdTech's CASE Network 2), Illustrative Mathematics curriculum data, granular learning components co-developed with the Achievement Network (ANet), and the Student Achievement Partners (SAP) Coherence Map for Mathematics progressions.

The server exposes three tools focused on standards work: looking up the authoritative text and metadata for any standard code (for example, 5.NBT.A.2 or RL.4.1), decomposing a standard into the discrete, teachable skills that make it up, and walking backward to prerequisites or forward to follow-on standards. This lets a model ground curriculum, lesson, and assessment work in real, citable standards data instead of hallucinated references.

Learning Commons is an initiative from the Chan Zuckerberg Initiative (CZI) focused on AI infrastructure for education. The MCP server is currently labeled an early access feature, meaning capabilities and schemas may change. It is provider-hosted at https://kg.mcp.learningcommons.org/mcp and authenticated with an API key created through the Learning Commons Platform.

Tools

Prerequisites

• A Learning Commons Platform account at platform.learningcommons.org
• A Knowledge Graph API key created in that platform
• An MCP-compatible client (Claude Desktop, Claude Code, Cursor, OpenAI Responses API, etc.)

Server details

• Server URL: https://kg.mcp.learningcommons.org/mcp
• Transport: Remote HTTP MCP server
• Auth header (either accepted):
  • x-api-key: YOUR_API_KEY
  • Authorization: Bearer YOUR_API_KEY

Claude Desktop / Claude Code / Cursor config

Add the server to your MCP client config (for example claude_desktop_config.json or .mcp.json):

{
  "mcpServers": {
    "learning-commons-kg": {
      "url": "https://kg.mcp.learningcommons.org/mcp",
      "headers": {
        "x-api-key": "YOUR_API_KEY"
      }
    }
  }
}

OpenAI Responses API config

{
  "type": "mcp",
  "server_label": "learning-commons-kg",
  "server_url": "https://kg.mcp.learningcommons.org/mcp",
  "require_approval": "never",
  "headers": {
    "x-api-key": "YOUR_API_KEY"
  }
}

Notes

• The MCP server is in early access. Tool names and schemas may change as capabilities expand.
• For broader Claude integration (chat-only, no developer setup), Learning Commons also offers a Claude connector via claude.ai/directory.

• Ground lesson planning in authoritative standards by resolving codes like 5.NBT.A.2 to their official statement and metadata before generating activities.
• Break a target Common Core math or ELA standard into granular learning components so an AI can scaffold an instructional sequence at the skill level.
• Trace prerequisite standards backward from a struggling student's current target to identify gaps for remediation.
• Plan forward curriculum sequences by walking the SAP Coherence Map progression from a mastered standard to its follow-on standards.
• Align AI-generated assessment items, rubrics, or worksheets to verified state or Common Core standards instead of hallucinated codes.

• "Look up the official statement for Common Core standard 5.NBT.A.2 and summarize what students need to know."
• "Decompose 8.F.B.4 into its learning components and draft a 3-lesson mini-unit covering each."
• "What are the prerequisite standards for RL.4.1? Use the Knowledge Graph to walk backward in the progression."
• "Find the Maryland version of RI.5.2, then list the next standards a student should target after mastering it."
• "I'm writing a 6th grade math diagnostic. Pull the learning components for 6.RP.A.3 and propose one item per component."

• Official server from Learning Commons (CZI), with data sourced from authoritative providers like 1EdTech CASE Network 2, Illustrative Mathematics, ANet, and Student Achievement Partners.
• Provider-hosted remote MCP server, so no install or local runtime is required.
• Covers Common Core plus 18 state standard sets, with both standards decomposition and progression mapping in one place.
• Accepts both x-api-key and Authorization: Bearer headers, which works cleanly across MCP clients and the OpenAI Responses API.

• Early access: tools, schemas, and behavior may change without notice.
• Progression data is currently limited to mathematics (SAP Coherence Map); ELA progression coverage is not exposed via the progressions tool.
• Only three tools today; no write operations, no curriculum or assessment authoring endpoints.

• Learning Commons Knowledge Graph REST API for non-MCP integrations or local querying via downloadable JSONL files.
• 1EdTech CASE Network 2 directly, for raw standards data without the MCP and learning components layer.
• Common Standards Project API for free programmatic access to Common Core and state standards (no MCP server, REST only).