SkilleoOpen source AI infrastructure and agent runtime

pip install skilleo — call a stable /v1 contract on Skilleo Cloud at api.skilleo.org. GPU bills stay with your provider.

Quickstart

Open source and Cloud

API reference

Same contract, two deployments

Open source SDK, capability spec, and provider interfaces. Cloud hosts authentication, execution, and billing at https://api.skilleo.org.

Provider routing

LLM — OpenAI, Claude, Gemini, MiniMax, AWS, or RunPod (any connected). Voice — MiniMax or ElevenLabs only. GPU — RunPod only.

Provider connections

client.providers.connect registers a prv_ pointer only — customer keys never stored by Skilleo. Each vendor bills the customer directly.

Async jobs

POST /v1/jobs returns 202. Poll, then download output.

Metering

Usage separates Skilleo subscription from provider_usage. No prepaid GPU wallets.

Python SDK

from skilleo import Client — HTTPS to api.skilleo.org/v1. Never imports RunPod.

Call Cloud in four lines

bash

pip install skilleo

python

from skilleo import Client

client, signup = Client.signup("dev@example.com")
client.providers.connect("runpod")  # you set RUNPOD_API_KEY in your secret store
run = client.capabilities.run("research.gather", version="1", input={"topic": "Kenya"})
print(client.runs.wait(run["id"])["data"]["output"])

Default base URL is https://api.skilleo.org. Override only if you run a compatible /v1 server yourself.

Open source vs Skilleo Cloud

Open source (`pip install skilleo`)	Skilleo Cloud
Python SDK	Hosted control plane
Capability specification	Authentication
Provider interfaces (RunPod GPU; MiniMax/ElevenLabs voice)	Subscriptions and billing
Agent SDK and examples	Managed execution
CLI and local runtime	Production orchestration

Same /v1 either way. Quickstart · Open source and Cloud