- CollectionAgent Runtime & Infrastructure
- Classificationagent-native
- MCPUnknown
- Agent SkillUnavailable
- InterfacesCLI · SDK · http
What it does
Modal is serverless infrastructure for AI workloads: inference, fine-tuning, batch jobs, notebooks, and dynamic sandboxes with sub-second cold starts, elastic GPU/CPU scaling, and unified observability. It is explicitly marketed for LLM inference, evals, RL environments, and MCP servers at scale — workloads that autonomous agents and their orchestrators spawn unpredictably.
Agent-facing interfaces
- Catalog protocols: CLI, SDK, http
- MCP: ⚠️ Not an MCP product; teams deploy MCP servers on Modal (see customer quotes on modal.com).
- Agent Skill: ⚠️ No official AgentSkills bundle located.
- Onboarding type:
sdk
How agents onboard
pip install modal
modal setup
# Define Modal app in Python per https://modal.com/docs/guide
See the full onboarding section in the dossier.
Protocol surface
| Interface | Detail |
|---|---|
| Python SDK | pip install modal |
| CLI | modal — deploy, run, logs |
| HTTP | Function endpoints — see docs |