- CollectionAgent Runtime & Infrastructure
- Classificationagent-native
- MCPAvailable
- Agent SkillUnavailable
- InterfacesMCP · REST
What it does
Scrapybara hosts remote desktops for computer-use agents (OpenAI CUA and similar): spin up Ubuntu, browser-only, or Windows VMs in milliseconds, stream the desktop, run bash, file edits, and browser automation via Playwright. It supports authenticated browser state (save/load), pause/resume, and hundreds of concurrent instances for parallel agent workloads.
Agent-facing interfaces
- Interaction pattern: Python/TypeScript SDK + Act SDK + MCP
- Catalog protocols: MCP, REST
- MCP: ✅ Community/org MCP repo
- Agent Skill: ⚠️ No official ClawHub skill pack documented as canonical.
- Onboarding type:
mcp
How agents onboard
Interaction pattern: Python/TypeScript SDK + Act SDK + MCP
pip install scrapybara
# or: npm install scrapybara
See the full onboarding section in the dossier.
Protocol surface
| Interface | Detail |
|---|---|
| Python | scrapybara on PyPI |
| TypeScript | scrapybara on npm |
| REST | Per docs |
| MCP | scrapybara-mcp repo |