edu-agentsSAFE
Open-source Agent Skills that help K-12 teachers plan lessons and build materials: lesson plans, slides, worksheets, quizzes, rubrics, differentiation.
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
Agent Skills that help K-12 teachers plan lessons and make teaching materials: objectives, unit and lesson plans, slide outlines, worksheets, quizzes with answer keys, rubrics, differentiation, feedback on student work, and messages to families.
Each skill is a folder with a SKILL.md file in the open Agent Skills format. Claude Code, Codex, Cursor, Grok, OpenCode, and other agents that read that format can load them. You install the skills once, then ask your agent in plain English: "Plan a 5-lesson Grade 8 unit on chemical reactions."
The skills are standards-agnostic. They work with Common Core, NGSS, a state or national curriculum, or your own goals, and they never invent standard codes.
Quick start
- Clone the repository:
git clone https://github.com/EduAgentsKit/edu-agents.git
- Copy or link the
skills/folders into the skills folder your agent reads. For Claude Code on macOS or Linux:
mkdir -p ~/.claude/skills cp -R edu-agents/skills/* ~/.claude/skills/
docs/install.md lists the folder for each agent, with links to each tool's documentation. You don't need Node.js or any other software to use the skills.
- Start your agent and ask for what you need:
Write a 45-minute Grade 3 lesson on unit fractions, with an exit ticket. Standard: CCSS 3.NF.A.1.
The agent picks the matching skills, asks for anything missing, and drafts the lesson.
Skills
93c4c9b0f60fOBSERVED · 2026-10-13Install
Commands as the repository documents them. They are shown, not run.
git clone https://github.com/EduAgentsKit/edu-agents.git
git clone https://github.com/EduAgentsKit/edu-agents.git
git clone https://github.com/EduAgentsKit/edu-agents.git
Host compatibility
What the documentation claims. We have not run a compatibility test.
| Host | Status | Notes |
|---|---|---|
| claude-code | mentioned | |
| codex | mentioned | |
| cursor | mentioned |
What it tells the agent
The instruction file, verbatim from the audited commit — this is the text the model reads, and the surface the audit's instruction layer examines. Quoted here so you can judge it without cloning anything.
--- name: edu-assessment-builder description: "Writes quizzes and tests with a blueprint, well-formed multiple choice and constructed response items across DOK levels, and an answer key with rationales. Use when a teacher asks for a quiz, test, unit assessment, review game questions, or an answer key." license: MIT --- # Assessment builder Write an assessment that measures the objectives and nothing else. ## Ask first 1. Grade, subject, and the objectives or unit plan to cover. 2. Length in minutes and number of items. 3. Item types allowed: multiple choice, short answer, extended response, performance. 4. Any accommodations the teacher already knows the class needs, such as read-aloud or extended time. ## Blueprint first Before writing items, show a blueprint table: objective, number of items, DOK levels, points. Aim for at least a quarter of the points at DOK 2 or higher, and include DOK 3 for a unit test. Get the teacher's OK on the blueprint when the assessment is longer than 10 items. ## Write items Follow [the item-writing rules](references/item-writing.md). The short version: - One clear problem per item. Put the question in the stem, not in the options. - Multiple choice gets one correct answer and three plausible distractors, each based on a real misconception. - No "all of the above", no "none of the above", no trick wording, no negative stems unless the word NOT is bold and capitalized. - Keep option lengths similar and vary the position of the correct answer. - Constructed response items name exactly what a full answer includes. ## Answer key For every item give: the correct answer, the objective, the DOK level, points, and a one-line rationale. For multiple choice, name the misconception behind each distractor. For constructed response, give a scoring guide or point to a rubric from the `edu-rubric-builder` skill. ## Output 1. Blueprint table. 2. Student copy with directions and point values. 3. Answer key. ## Guardrails - Standards: use only the codes the teacher gives you. If there are none, write `[standard: ask teacher]`. Never invent a code or quote standard text from memory. - Privacy: work without student names or identifying details. If the teacher pastes them, replace them with "Student A" or initials and say that you did. Ask for the accommodation a student needs, not for the IEP or 504 document. - Age and safety: fit reading level, content, and examples to the grade. Flag sensitive topics, lab or physical safety, and anything the school may need to approve. - Accuracy: work every answer key yourself. Mark any fact you are not sure of with `[verify]`. - The teacher decides: you draft. Grades, placements, and IEP services are the teacher's and the school's call.
Trust audit
SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | PASS |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | PASS |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- none-observed
- Network
- none-observed
- Shell
- none-observed
- Dependencies
- not all pinned
- Secrets in source
- none-found
Findings (1)
yaml
Gates applied: no_behavioural_pass.
93c4c9b0f60ffull audit observations/trust-audit/skill/eduagentskit__edu-agents.json · Report an issue / request a re-scanAudit history
Every audit this skill has had.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-13 | 93c4c9b0f60f | SAFE | B | 89 | first audit |
Questions
What does the edu-agents skill do?
Open-source Agent Skills that help K-12 teachers plan lessons and build materials: lesson plans, slides, worksheets, quizzes, rubrics, differentiation.
Is edu-agents safe to install?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean skill reads B.
What can edu-agents access on my machine?
The audit observed no filesystem, network or shell use at all in its source.
Which assistants does edu-agents work with?
Its documentation mentions claude-code, codex and cursor. That is what the text claims, not a compatibility test we ran.
How current is this page?
The grade is for one exact copy of the source (93c4c9b0f60f), read on 2026-10-13. The repository is watched, and a new audit runs when it changes — this is the first audit.