EmbodiFlowDocs
Get startedDataWorkflows and algorithmsTraining and deploymentDeveloper reference
  • Developer reference

Authentication

  • Authentication overview
  • API keys
  • Personal access tokens
  • Browser sessions

REST API

  • Using the REST API
  • All endpoints
  • Projects
    • Recordings
      • Upload
        • Episodes and versions
          • Annotation
            • Quality and tasks
              • Search
                • Workflows
                  • Algorithms
                    • Models
                      • Artifacts and export
                        • Billing and usage
                          • Team and governance
                            • Account and notifications
                              • Webhook endpoints

                                Events

                                • Outbound events
                                • Signature verification

                                Tools

                                • CLI
                                • SDK
                                • MCP

                                Updates

                                • Changelog

                                © 2026 EmbodiFlow

                                llms.txtOpenAPI JSON
                                1. Developer reference
                                2. Tools
                                3. SDK

                                SDK

                                TypeScript and Python clients, generated from the public OpenAPI document.

                                If you're writing code against EmbodiFlow, an SDK saves you the plumbing. Both clients cover every public /v1 operation and sign in with a personal access token. They also handle the polite parts for you: on 429 they wait for Retry-After and retry, and they reuse idempotency keys so a retried write isn't applied twice.

                                TypeScript

                                The package is @embodiflow/sdk. It's published to the same GitHub Packages registry as the CLI, so you need read access to that registry to install it.

                                import { EmbodiFlowClient, paginate } from '@embodiflow/sdk';
                                
                                const client = new EmbodiFlowClient({
                                  baseUrl: 'https://api.embodiflow.io',
                                  apiKey: process.env.EMBODIFLOW_PAT!,
                                });
                                
                                for await (const project of paginate((pageToken) =>
                                  client.api.projects.list({ query: { pageSize: 25, pageToken } }),
                                )) {
                                  console.log(project.id);
                                }

                                Python

                                The distribution is embodiflow-sdk, and you import it as embodiflow. It isn't on PyPI; instead, each deployment serves its own build as a pip index:

                                pip install --index-url https://api.embodiflow.io/v1/sdk/simple/ embodiflow-sdk

                                The Python package doesn't install a command. For the terminal, use the CLI.

                                Writing an algorithm is a different thing: algorithms use the embodiflow module that the algorithm runtime provides. See the algorithm SDK reference.
                                Previous

                                CLI

                                Next

                                Changelog