datashift-io/mcp-server
Official MCP server for Datashift — human-in-the-loop review workflows for AI agents
Platform-specific configuration:
{
"mcpServers": {
"mcp-server": {
"command": "npx",
"args": [
"-y",
"mcp-server"
]
}
}
}Add the config above to .claude/settings.json under the mcpServers key.
[](https://opensource.org/licenses/MIT)
Model Context Protocol server for Datashift. Allows AI agents to submit tasks for human or AI review and receive decisions via MCP tools.
Datashift adds human review checkpoints to AI agent workflows. This MCP server enables agent-initiated review — the agent decides when to request human oversight based on its own reasoning.
| Tool | Description | |------|-------------| | submit_task | Submit a task for review (approval, classification, scoring, labeling, augmentation) | | get_task | Check task status and retrieve review results | | list_queues | List available review queues | | get_queue | Get queue configuration and review options |
| Resource | Description | |----------|-------------| | task:///{task_id} | Access task details and review status |
npm installCopy .env.example to .env and configure your environment variables.
npm run build
npm startUses OAuth 2.0 with PKCE for secure agent authentication. AI agents obtain a JWT from the Datashift API's OAuth endpoints, which is validated by the MCP server and passed through to API calls.
Loading reviews...