Claw-ED
A local-first AI teaching assistant that drafts lessons, student handouts and slides from your own lesson plans.
Import your old lesson plans, then generate teacher and student DOCX plus PPTX slides with your chosen model.
Claw-ED is a Python command-line assistant for a single teacher. You feed it your existing PDFs, DOCX and PPTX files and it drafts new lessons that follow your style; it is a beta, so you review everything it produces.
What it is
Claw-ED extracts text from the files you import, indexes it for retrieval and builds a teaching-style profile. The lesson bundle tool then produces a teacher DOCX, a student DOCX and PPTX slides, with optional assessments and differentiated materials. A local dashboard and Telegram bot are included.
Who it's for
It suits a teacher comfortable with a terminal and pip who has a library of lesson materials to build on. It does not suit a school wanting a multi-user platform: the README says it is intended for one teacher per instance and student-facing features are experimental.
Key features
Import your own materials
Ingests PDF, DOCX, PPTX, TXT and Markdown folders, then searches them when drafting.
Teaching-style profile
Builds a profile from your files so drafts follow how you already write lessons.
Lesson bundle export
One tool creates a teacher DOCX, student DOCX and PPTX slides you can edit in Word or PowerPoint.
Recoverable lesson jobs
A queue with cancel, recover and resume commands keeps completed generation phases after a restart.
Source manifest
Flags quotations needing review so you can check what the model pulled from your sources.
Choose your model
Adapters for Anthropic, OpenAI, Google Gemini, Ollama and OpenRouter, including local models.
Up and running in 3 steps
Our honest verdict
What's great
- It starts from your own lesson plans and slides, so drafts reflect your material instead of a generic template.
- Output is plain DOCX and PPTX that you edit in tools you already use.
- You can run a local model through Ollama and keep prompts off cloud providers, if you avoid network features.
- Job queue commands (cancel, recover, resume) save partial lesson bundles instead of losing a long generation.
Where it falls short
- It is marked Beta and needs Python 3.11 or newer, a virtual environment and command-line use; there is no installer.
- Model usage is not included: you pay your chosen provider's API costs or run a local model on your own hardware.
- The README warns that local storage does not mean all processing stays on the device, since hosted models receive your prompts and selected content.
- Single-teacher design; student-facing games and live sessions are experimental and some session state does not survive restarts.
Pricing & license
Claw-ED is MIT-licensed and free to install. Model usage is not included, so the cost is your chosen provider's API fees, or hardware if you run a local model. The README says supporting the developer is optional.
Open remote AI-training roles
767 open remote AI-training roles, updated November 14, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Alternatives
These reviewed AI tools do the same job instantly — from our scored directory.
Use it, or have us set it up
FAQ
Does Claw-ED send my files to the cloud?
Files and settings are stored locally under ~/.eduagent/, but hosted models receive the prompts and content needed for each request. For local-only use, configure a local model and avoid web search and integrations.
Which models does it work with?
Anthropic, OpenAI, Google Gemini, Ollama and OpenRouter. The README suggests trying Qwen 3.5 9B locally as a starting point and says results vary by provider.
Can a whole school use it?
The README says it is intended for a single teacher per instance and is not a multi-tenant school platform.
