Runcell
AI agent for Jupyter notebooks that writes, runs, and debugs code automatically
About Runcell
Runcell is an AI-powered agent for Jupyter notebooks that automates writing Python code, executing cells, debugging errors, and explaining data analysis results in real time. It transforms Jupyter into a modern AI IDE with four modes: interactive learning with live examples, autonomous agent mode that plans and executes tasks end-to-end, smart edit mode for step-by-step reasoning, and AI-enhanced Jupyter for natural language queries. Uniquely, Runcell can understand visualizations and image outputs from your code, not just text. It works as a lightweight browser extension — no desktop app downloads, new IDE, or API key setup needed.
Pros & Cons
Pros
- Works inside existing Jupyter — no new IDE or tool switching needed
- Only AI agent that truly understands visual outputs (charts, plots)
- Free tier with 50 credits per month for casual use
- Autonomous mode handles end-to-end tasks including error recovery
- One-click install with no API key setup required
- Context-aware: understands your full notebook state and variables
Cons
- Limited to Jupyter Lab ecosystem — no VS Code or standalone support
- Credit-based system limits heavy usage on free and Pro tiers
- Relatively new product — smaller community than GitHub Copilot
- Pro plan at $20/month is comparable to full AI coding assistants
- Best suited for data science workflows rather than general coding
Key Features
Autonomous Agent Mode
Plans tasks, writes code, runs cells, fixes errors, installs packages, and retries until the goal is complete
Visual Output Understanding
Uniquely understands charts, plots, and image outputs from your code — not just text results
Interactive Learning Mode
AI teacher that explains concepts with live runnable examples directly in your notebook
Smart Edit Mode
Advanced reasoning mode that analyzes your notebook deeply for reliable refactors and step-by-step solutions
Context-Aware AI
Works with existing variables, prior cell outputs, and charts so follow-up requests stay grounded in kernel state
Auto Execution & Debugging
Runs cells safely, diagnoses errors automatically, and suggests fixes without manual intervention
Natural Language Queries
Generate charts, functions, and reports from plain English prompts inserted directly into your notebook
Jupyter Native
Lightweight extension that works in your existing Jupyter Lab — no new IDE, desktop app, or API key setup required
Pricing
- 50 credits per month
- 50 bonus credits during beta
- AI code execution
- Community support
- No credit card required
- 500 credits per month
- Unlimited code completions
- Priority support
- All AI modes included
- 1,500 credits per month
- Team management dashboard
- Access to all models
- Priority access to new features
- Privacy mode