SkillCommunity★ 380

Education Agent Skills

Updated Jul 7, 2026

Education Agent Skills by GarethManning is an open-source library of 165 skills across 20 domains grounded in named educational research. Covers memory science, metacognition, curriculum design, assessment, EAL/D, Montessori, systems thinking, inclusive design, and a student-facing domain. Installs into Claude Code, Codex, Hermes Agent, and any Agent Skills-compatible tool.

View on GitHub

What it does

claude plugin install https://github.com/GarethManning/education-agent-skills
Install all 165 education skills into Claude Code CLI.
spaced-practice-scheduler skill (skills/memory-learning-science/)
Generates a week-by-week spaced retrieval schedule for a topic list using Cepeda et al. (2006) optimal spacing intervals.
backwards-design skill (skills/curriculum-design-assessment/)
Produces a backwards-design unit plan starting from desired outcomes and working back to assessment evidence and learning activities.
student-facing skills (Domain 20, skills/student-learning/)
13 live-session skills including retrieve-first gate, progressive hint ladder, confidence calibration check, and fading manager — for AI tutoring flows that run during a student's study session.

Files (1)

OVERVIEW.mdprimary · markdown · 3.0 KB

## What it does

Education Agent Skills by GarethManning is an open-source library of 165 evidence-based pedagogical skills across 20 domains, built for Claude Code, OpenAI Codex, Hermes Agent, and any Agent Skills-compatible tool. Every skill is grounded in named research (specific authors, studies, findings) and rated by evidence strength (Strong, Moderate, Emerging, Original). Frameworks without empirical support — learning styles, VAK — are excluded; EXCLUSIONS.md documents why.

The 20 domains include: Memory & Learning Science (8 skills: retrieval practice, spacing, interleaving, cognitive load, dual coding, elaborative interrogation, feedback), Self-Regulated Learning & Metacognition (5 skills), Explicit & Direct Instruction (5 skills), Questioning & Dialogue (5 skills), Literacy/Writing/Critical Thinking (7 skills), EAL/D & Language Development (5 skills), Curriculum Design & Assessment (13 skills: backwards design, rubric generation, formative assessment, differentiation, gap analysis, learning progressions, PBL), Wellbeing/Motivation/Agency (12 skills), Professional Learning & Teacher Development (10 skills), Global & Cross-Cultural Pedagogies (9 skills), Environmental & Experiential Learning (6 skills), AI Learning Science (14 skills), AI Literacy (7 skills), Montessori & Alternative Evidence-Based Approaches (4 skills), Original Frameworks (17 skills), Curriculum Alignment (4 skills), Historical Thinking (10 skills), Systems Thinking (8 skills), Inclusive Design (3 skills), and Student-Facing Learning Skills (13 live-session skills for learners).

Each skill is a folder under skills/ containing SKILL.md with YAML frontmatter (typed inputs/outputs, evidence sources, chaining metadata, evidence strength rating). No build step — copy skill folders into any tool's skills directory.

## Install

Claude CoWork (easiest): Customize → (+) Add Plugin → paste https://github.com/GarethManning/education-agent-skills

Claude Code CLI:

    claude plugin install https://github.com/GarethManning/education-agent-skills

Codex (local):

    git clone https://github.com/GarethManning/education-agent-skills.git && cd education-agent-skills && codex plugin marketplace add "$PWD"

Hosted MCP (requires free auth token from the repo's signup form):

    { "mcpServers": { "education-skills": { "type": "streamable-http", "url": "https://mcp-server-sigma-sooty.vercel.app/mcp", "headers": { "Authorization": "Bearer <token>" } } } }

Manual: copy skill folders from skills/ into your tool's skills directory.

## When to use

Use this when you need lesson planning, rubric generation, formative assessment, or curriculum design that is grounded in actual educational research, not generic instructional templates. The spaced-practice-scheduler skill, for example, follows Cepeda et al.'s (2006) meta-analysis on optimal spacing intervals and returns a week-by-week schedule with interleaved retrieval activities. Use the student-facing domain (Domain 20) when building AI tutoring flows that adapt to learner progress in real time.