- CollectionCode Execution
- Classificationagent-native
- MCPAvailable
- Agent SkillUnavailable
- InterfacesMCP · REST · CLI · SDK · http · API
What it does
Runloop provides Devboxes: secure, isolated sandboxes for running AI-generated code and long-running agent workflows, with fast startup, optional suspend/resume, disk snapshots and branching for agent state, repo-aware environments, secrets/objects stores, and benchmark jobs to evaluate coding agents (e.g., SWE-Bench-style runs). The surface area is explicitly marketed for AI agents and AI software engineering workloads—not generic batch VMs only.
Agent-facing interfaces
- Interaction pattern: SDK / REST + CLI + MCP
- Catalog protocols: MCP, REST, CLI, SDK, http, API
- MCP: ✅ Available (via official CLI)
- Agent Skill: ⚠️ No official npx skills add … pack from Runloop yet.
- Onboarding type:
mcp
How agents onboard
Interaction pattern: SDK / REST + CLI + MCP
pip install runloop-api-client
# or: npm install @runloop/api-client (see GitHub api-client-ts)
export RUNLOOP_API_KEY=...
See the full onboarding section in the dossier.
Protocol surface
| Interface | Detail |
|---|---|
| Python SDK | pip install runloop-api-client — api-client-python |
| TypeScript SDK | @runloop/api-client — api-client-ts |
| REST API | Platform API per docs.runloop.ai |
| CLI | npm install -g @runloop/rl-cli → rli |
| MCP | rli mcp start / rli mcp install |