Platform availability

Skilleo Cloud runs at https://api.skilleo.org. Install the SDK with pip install skilleo and authenticate with a sk_live_… key from signup.

API surface

Area	Available on Cloud
Account and API keys	Yes
Provider connections	Yes — Secrets
Versioned capabilities	Yes — Capabilities
Runs and jobs	Yes
Compute, storage, templates, endpoints, catalog	Yes
Artifacts, workflows, plans, tasks, agents	Yes
Usage and billing	Yes
Open source Python SDK	Yes — talks to /v1 over HTTPS

Self-hosted deployments that implement the same /v1 contract can be used with a custom base URL. See Open source and Cloud.

Capabilities

Capability	Version	Requires provider connection
research.gather	1	No (optional LLM for synthesis)
llm.generate, text.generate	1	Yes — any connected LLM
speech.generate, voice.clone	1	Yes — MiniMax or ElevenLabs
image.generate	1	Yes — RunPod
composition.layout	1	No

Discover the live registry:

curl -s https://api.skilleo.org/v1/capabilities \
  -H "Authorization: Bearer $SKILLEO_API_KEY"

Each capability includes source.routing (allowed connections, default provider) and source.providers (backends you can connect).

Provider routing

Workload	Allowed connections	Rule
llm.generate	OpenAI, Anthropic, Gemini, MiniMax, AWS, RunPod	Use any connection you register
speech.generate, voice.clone	MiniMax, ElevenLabs	Connect one or both
image.generate, GPU infra	RunPod	RunPod only

Default LLM and voice provider when you connect several: MiniMax (default_provider in the capability registry). Override with options.provider on runs.

Supported provider connections

Register connections with POST /v1/providers/connections. Skilleo stores a pointer (prv_…) to your secret — never the key itself.

Connection	Billed by	Used for
openai	OpenAI	LLM
anthropic	Anthropic	LLM
gemini	Google	LLM
minimax	MiniMax	LLM, speech, voice clone
elevenlabs	ElevenLabs	Speech, voice clone
aws	AWS	LLM (Bedrock)
runpod	RunPod	GPU, image generation, optional LLM

Skilleo subscription covers platform access. Provider usage is billed by each vendor directly.

Verify your setup

After connecting a provider, confirm routing with a short run:

# LLM (requires a connected LLM provider)
curl -s -X POST https://api.skilleo.org/v1/runs \
  -H "Authorization: Bearer $SKILLEO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"target":"capability","name":"llm.generate","version":"1","input":{"objective":"Reply with one sentence about Kenya wildlife."},"options":{"optimize_for":"cost"}}'

# Poll until completed
curl -s https://api.skilleo.org/v1/runs/run_... \
  -H "Authorization: Bearer $SKILLEO_API_KEY"

# Speech (requires MiniMax or ElevenLabs)
curl -s -X POST https://api.skilleo.org/v1/runs \
  -H "Authorization: Bearer $SKILLEO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"target":"capability","name":"speech.generate","version":"1","input":{"text":"Hello from Skilleo."}}'

Research-only workloads work without a provider connection:

curl -s -X POST https://api.skilleo.org/v1/runs \
  -H "Authorization: Bearer $SKILLEO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"target":"capability","name":"research.gather","version":"1","input":{"topic":"Kenya wildlife"}}'

If a capability needs a provider you have not connected, the API returns 400 CREDENTIALS_MISSING.

Roadmap

The /v1 contract is stable for application development. Upcoming areas include agent-targeted runs on /v1/runs, marketplace apps, and expanded self-hosted tooling. Capabilities marked planned in the registry are documented before general availability.