You want to use AI coding assistants to help write code, but:
- You don't want AI touching your local machine or production systems
- You need different AI providers for different tasks (fast inference vs high quality)
- You want AI to work in clean, reproducible environments
- Setting up AI tools manually is tedious and error-prone
Runcell solves this by letting you ask Claude (in your IDE) to spin up isolated workspaces where AI assistants are ready to use immediately.
You chat with Claude Code in VS Code. Claude has access to Runcell's MCP tools.
You: "Create a Go workspace and use Cline to build me a REST API with user CRUD"
Claude: I'll create an isolated workspace with Cline pre-configured, then
delegate the task to the inner AI.
[Creates workspace with Cerebras-powered Cline]
[Runs: cline -y "Create a REST API with user CRUD endpoints"]
[Reports results back to you]
You: "Add authentication to it"
Claude: [Runs: cline -y "Add JWT authentication to the API"]
Done! I've added JWT auth. Here's what changed...
You never leave your IDE. You never configure API keys manually. You never touch a terminal.
1. Open terminal
2. Create project directory
3. Install cline globally
4. Run: cline auth --provider openai --apikey sk-xxx --modelid gpt-4 --baseurl https://...
5. Hope you remembered all the flags
6. Hope you didn't expose your API key in shell history
7. Start working, realize you need a different model
8. Reconfigure everything
9. Repeat for every new project
You: Create a Python workspace with Cline configured for fast inference,
then build me a CLI tool that converts CSV to JSON.
Claude: I'll set that up for you.
✓ Created workspace "csv-tool" with Python 3.12
✓ Cline configured with Cerebras (qwen-3-32b)
✓ Running inner AI to build the CLI tool...
Done! The inner AI created:
- cli.py (main entry point with argparse)
- converter.py (CSV parsing and JSON output)
- tests/test_converter.py (unit tests)
All tests pass. Want me to show you the code or make changes?
When you ask Claude to create a workspace and run an inner AI:
- Claude receives your request in VS Code
- Claude calls Runcell MCP tools to create an isolated Kubernetes pod
- The pod starts with AI pre-configured (Cline, Claude Code, etc.)
- Claude runs the inner AI with your prompt
- Claude reports results back to you in the chat
┌─────────────────────────────────────────────────────────────┐
│ VS Code │
│ ┌───────────────────────────────────────────────────────┐ │
│ │ Claude Code Chat │ │
│ │ │ │
│ │ You: Build me a web scraper in Go │ │
│ │ │ │
│ │ Claude: I'll create a workspace and have the inner │ │
│ │ AI build that for you. │ │
│ │ │ │
│ │ ✓ Workspace created (ws-a1b2c3) │ │
│ │ ✓ Cline configured with Cerebras │ │
│ │ ⟳ Inner AI working... │ │
│ │ │ │
│ │ Done! Created: │ │
│ │ • main.go - CLI entry point │ │
│ │ • scraper/scraper.go - Core scraping logic │ │
│ │ • scraper/parser.go - HTML parsing │ │
│ │ │ │
│ │ You: Add rate limiting │ │
│ │ │ │
│ │ Claude: ⟳ Asking inner AI to add rate limiting... │ │
│ │ │ │
│ └───────────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────────────┐
│ Kubernetes (invisible to you) │
│ │
│ Pod: ws-a1b2c3 │
│ ├── Go 1.22 runtime │
│ ├── Cline (pre-configured with Cerebras API) │
│ └── /workspace (isolated filesystem) │
│ ├── main.go │
│ └── scraper/ │
│ ├── scraper.go │
│ └── parser.go │
└─────────────────────────────────────────────────────────────┘
No terminal. No manual configuration. Just chat with Claude.
Each workspace is a separate container. AI can't access your local files, credentials, or other projects.
Claude sets up the inner AI automatically. You just describe what you want.
API keys are managed by Runcell and injected securely. Never exposed in chat or shell history.
Claude picks the right AI for the job, or you can specify:
- "Use fast inference" → Cerebras
- "Use Claude for this" → Anthropic direct
- "Use the cheap option" → OpenRouter
You: Create a workspace and build me a REST API in Go with
health check and user CRUD. Use fast inference.
Claude: [Creates workspace, configures Cline with Cerebras, runs task]
Done in 45 seconds! Here's what was created...
You: In the csv-tool workspace, have the inner AI review
converter.py for potential bugs and edge cases.
Claude: [Runs: cline -y "Review converter.py for bugs and edge cases"]
The inner AI found 3 issues:
1. No handling for empty CSV files
2. Unicode characters may cause issues
3. Memory usage could be high for large files
Want me to have it fix these?
You: Add support for YAML output too
Claude: [Runs: cline -y "Add YAML output support to the converter"]
Added! New files:
- formats/yaml.py
Updated:
- cli.py (new --format yaml flag)
- tests/test_converter.py (YAML tests)
You: The JSON output has extra whitespace. Fix it.
Claude: [Runs: cline -y "Fix: JSON output has extra whitespace"]
Fixed! The issue was in converter.py line 42 -
was using indent=4 instead of indent=None for compact output.
The workspace images come with these pre-installed:
| Tool | What it does |
|---|---|
| Cline | General-purpose AI coding assistant (works with any provider) |
| Claude Code | Anthropic's official assistant |
| OpenCode | Open-source alternative |
| Auggie | Augment Code's assistant |
Claude (outer) picks which one to use based on your request, or you can specify.
Q: Do I need to know Kubernetes? A: No. You just chat with Claude. Kubernetes is invisible to you.
Q: Do I need to configure API keys? A: Once, in Runcell's credential store. After that, Claude handles injection automatically.
Q: Can I see what the inner AI is doing? A: Yes, Claude shows you the output and what files were created/changed.
Q: What if the inner AI makes a mistake? A: Just tell Claude to fix it. Or delete the workspace and start fresh - it's isolated.
Q: Is my code safe? A: Yes. Each workspace is isolated. The inner AI only sees /workspace in its container.
Q: Can I download the code? A: Yes, ask Claude to copy files from the workspace, or use the file browser tools.
Runcell - Let Claude orchestrate AI coding assistants in isolated workspaces