Reference
For integrators: anyone who calls the platform from code or needs the exact value of a limit or a code.
Three APIs, three ways to authenticate
| API | Base path | Authentication | Who uses it |
|---|---|---|---|
| Agents and Workflows | /builder/v1 | Studio session or token (Authorization: Bearer) | Automate what you do in Studio; trigger published workflows by webhook. |
| Labs (Frida) | /workbenches/v1 | Studio session or token | Create and share labs from scripts. |
| InferNode | /v1 on the InferNode domain | sk-saptiva-... key (API > API Keys) | Call the models from any application, with the OpenAI SDK. |
| Credentials | See the installation's OpenAPI | An administrator's session | Rotate provider keys by script. |
The exact paths and schemas for Agents, Workflows and Credentials are in the OpenAPI that your installation serves at /builder/v1/openapi.json; this reference summarizes the contract and what changes between versions.
Lookup tables
- Limits and quotas: fixed Studio caps, InferNode limits per plan and lab limits per installation, retention.
- Measured performance: latency, output rate, concurrency and long context of the models, with the method to measure them again.
- Error codes: HTTP and application codes of the three APIs, plus the Studio messages and the page each one leads to.
- Glossary: one term per concept, with the name the interface uses.
- Release notes: visible changes per Studio version.