Python SDK for integrating human-in-the-loop checkpoints into AI agent workflows.
Datashift enables AI agents to submit tasks for human review before committing changes. Create review queues, define review types (approval, classification, labeling, scoring, augmentation), and integrate human oversight into any AI workflow — via REST API or MCP.
Coming soon. The Python SDK is under active development. In the meantime, you can use the REST API directly or the TypeScript SDK.
- Datashift Website — Product overview and sign up
- Documentation — Guides, API reference, and examples
- TypeScript SDK — Available now on npm
- MCP Server — Connect AI agents via Model Context Protocol
- Console — Manage queues, reviewers, and API keys
- GitHub — SDKs and sample projects
MIT