Agent Skills are open, ready-to-use skills that help AI assistants produce high-quality, standards-aligned K-12 teaching materials. Each skill packages the instructions, references, and guardrails an agent needs to reliably complete a teacher workflow — so the same task produces consistent, classroom-ready results.
The skills are developed to be grounded in learning science and leverage Learning Commons Knowledge Graph. They are built to be cross-platform and model-agnostic: usable with any agent runtime that supports the open skills format.
Current use cases include:
- Lesson planning: Builds classroom-ready, standards-aligned lesson plans, optionally aligned to a teacher's curriculum.
- Lesson differentiation: Adapts an existing lesson into tiered versions (below / at / above proficiency-level) and for specific student needs, keeping core content consistent across tiers.
Both of these initial use cases were co-developed with Anthropic. A companion K-12 teacher skills repository from Anthropic accompanies this work.
To get started, see Quick Start below and the skills/ directory for what each skill does.
| Path | Description |
|---|---|
| skills/ | The published agent skills and example prompts to use the skills |
| evals/ | Evaluation rubrics used to evaluate the quality of skill outputs |
| LICENSE | Open source license details |
This quick start guide walks you through getting set up with the Learning Commons skills and MCP server inside a coding agent. It's an easy install and a great way to explore how the skills work. You can also clone this repo and use these skills and the MCP server with an LLM directly through APIs — the route you'd take to build them into your own product experience.
Note for Claude Code: if your project doesn't have a
.claude/directory yet, runmkdir -p .claudebefore installing.
npx skills add learning-commons-org/agent-skillsThis installs the skills into any agent runtime that supports the open skills format (e.g. Claude Code, Cursor, Codex).
Note: Knowledge Graph datasets carry varying licenses. Some datasets aren’t readily accessible by default due to license requirements.
The skills are grounded in Learning Commons Knowledge Graph via its MCP server — not required, but strongly recommended for accurate, standards- and pedagogy-aligned output. Create an API key in the Learning Commons Platform, then add the server to your agent. For example, in Claude Code:
claude mcp add --transport http learning-commons-kg \
https://kg.mcp.learningcommons.org/mcp \
--header "Authorization: Bearer $LC_API_KEY"Prompt your agent with a typical teaching request — the matching skill loads automatically:
- "I need a lesson for tomorrow on rounding to the nearest hundred for my 3rd graders."
- "Differentiate this 6th grade food webs lesson for students below / at / and above proficiency level (find the lesson here: https://www.calacademy.org/educators/lesson-plans/how-stable-is-your-food-web)"
See skills/example-prompts.md for more examples.
Check the generated materials against the same rubrics we use to evaluate these skills — pedagogy, rigor, formatting, and model scaffolding. See evals/ for the rubrics and instructions on running them as an LLM-as-judge.
We want to hear from you. For questions or feedback, please open an issue or reach out to us at support@learningcommons.org.
Learn more about our work or partner with us to:
- Co-develop new skills for K-12 workflows
- Get early access to new skills, tools, and Knowledge Graph
- Receive personalized support from the Learning Commons team
Contact us here.
If you believe you have found a security issue, please responsibly disclose by contacting us at security@learningcommons.org.
The resources provided in this repository are made available "as-is", without warranties or guarantees of any kind. They may contain inaccuracies, limitations, or other constraints depending on the context of use. Use of Learning Commons services, including the MCP server, is subject to our Terms of Use.
By accessing or using these resources, you acknowledge that:
- You are responsible for evaluating their suitability for your specific use case.
- Learning Commons makes no representations about the accuracy, completeness, or fitness of these resources for any particular purpose.
- Any use of the materials is at your own risk, and Learning Commons is not liable for any direct or indirect consequences that may result.
Please refer to each resource's README, license, and associated docs for any additional limitations, attribution requirements, or guidance specific to that resource.