Skip to main content
The Admin API provides a comprehensive interface for managing Amp’s data pipeline operations. This API allows you to deploy datasets, trigger extraction jobs, monitor progress, manage distributed workers, and configure external data providers.

​
Base URL

The Admin API runs on port 1610 by default: 
http://localhost:1610

​
Key Capabilities

​
Dataset Management

Handle the lifecycle of data extraction configurations:
  • List all registered datasets from the metadata database
  • Register new dataset configurations with versioning support
  • Trigger data extraction jobs for specific datasets or versions
  • Retrieve dataset details including tables and storage locations

​
Job Control

Control and monitor data extraction and processing jobs:
  • List and retrieve job information with pagination
  • Trigger extraction jobs with optional end block configuration
  • Stop running jobs gracefully
  • Delete jobs in terminal states (Completed, Stopped, Failed)

​
Storage Management

Manage locations where dataset tables are stored:
  • Supports local filesystem, S3, GCS, and Azure Blob Storage
  • List storage locations and their associated files
  • Delete locations with comprehensive cleanup
  • Query file information including Parquet metadata

​
Provider Configuration

Configure external blockchain data sources:
  • Create, retrieve, and delete provider configurations
  • Support for EVM RPC endpoints and Firehose streams
  • Providers are reusable across multiple datasets

​
Worker Management

Monitor and manage distributed worker nodes:
  • List all registered workers
  • Get detailed worker information including version and build metadata
  • Track worker health via heartbeat timestamps

​
Authentication

The Admin API currently does not require authentication. Ensure you run it in a secure environment and restrict network access appropriately.

​
Error Responses

All endpoints return errors using a standard format: 
{
  "error_code": "DATASET_NOT_FOUND",
  "error_message": "dataset 'eth_mainnet' version '1.0.0' not found"
}
​
error_code
string
required
Machine-readable error code in SCREAMING_SNAKE_CASE format. Error codes are stable across API versions and should be used for programmatic error handling.
​
error_message
string
required
Human-readable error message providing detailed context. Messages may change over time - use error_code for programmatic decisions.

​
Common Patterns

​
Dataset Versioning

Datasets support multiple version reference types:
  • Semantic version (e.g., 1.2.3) - specific tagged version
  • Manifest hash - SHA256 hash of the dataset configuration
  • latest - highest semantic version
  • dev - development version tag

​
Pagination

List endpoints use cursor-based pagination:
  • Use limit query parameter to control page size (max 1000)
  • Use last_job_id or similar cursor parameter for subsequent pages
  • Response includes next_cursor field when more results are available

​
Idempotency

Many destructive operations are idempotent:
  • Deleting a non-existent resource returns success
  • Safe to retry failed requests without side effects
  • Helps build resilient automation

​
Next Steps

Dataset Management

Register and manage dataset configurations

Job Control

Trigger and monitor extraction jobs

Worker Management

Monitor distributed worker nodes

Provider Config

Configure blockchain data sources

Build docs developers (and LLMs) love

Get started for freeTalk to us