Goldsky Dataset Discovery

Explore and discover available blockchain datasets for Turbo pipelines.

Triggers

Invoke this skill when the user:

• •Asks "what data does Goldsky have?" or "what chains are supported?"
• •Wants to know if a specific chain or dataset is available
• •Needs help finding the right dataset name format
• •Says "help me find a dataset" or "what datasets are available?"
• •Is unsure which dataset to use for their pipeline
• •Asks about ERC-20 transfers, NFT data, logs, or other blockchain data types
• •Mentions /goldsky-datasets

Agent Instructions

IMPORTANT - Avoid goldsky dataset list: This command is slow (30-60+ seconds) and often times out. Use the reference tables below instead. Only run the command if you need to verify a specific dataset exists or find an exact version.

When this skill is invoked:

Step 1: Use the Reference Tables First

Do NOT run goldsky dataset list upfront. Instead:

1. •Check the Verified Dataset Reference table below for exact dataset names
2. •Use goldsky turbo validate on a test YAML to verify a dataset exists (fast, ~3 seconds)
3. •Only run goldsky dataset list | grep "specific_name" if you need version info

Step 2: Match User Need to Dataset

Based on what they're building, recommend directly from the reference:

Step 3: Validate Before Presenting

ALWAYS validate the dataset exists before giving it to the user:

# Quick validation test (fast, ~3 seconds)
goldsky turbo validate - <<'EOF'
name: test
resource_size: s
sources:
  test:
    type: dataset
    dataset_name: <chain>.<dataset_type>
    version: 1.0.0
    start_at: latest
transforms: {}
sinks:
  out:
    type: blackhole
    from: test
EOF

If validation fails with "Dataset not found", try alternate naming (see Verified Dataset Reference).

Dataset Reference Files

Detailed dataset and chain information is in the data/ folder.

Data location: data/ (relative to this skill's directory)

Quick Reference

Common Datasets

Important: Use raw_transactions, NOT transactions

Popular Chain Prefixes

See data/chain-prefixes.json for complete list with chain IDs.

Common Dataset Types

EVM Chains

Solana

Dataset Name Format

All datasets follow the pattern: <chain_prefix>.<dataset_type>

Examples:

• •ethereum.erc20_transfers - ERC-20 transfers on Ethereum mainnet
• •base.logs - All event logs on Base
• •matic.blocks - Block data on Polygon
• •solana.token_transfers - SPL token transfers on Solana

Finding Dataset Versions

Datasets are versioned. To find available versions:

goldsky dataset list | grep "base.erc20"

Common versions:

• •1.0.0 - Initial version
• •1.2.0 - Enhanced schema (common for ERC-20 transfers)

When in doubt, use the latest version shown in goldsky dataset list.

Common Discovery Patterns

"I want to track USDC transfers on Base"

1. •Dataset: base.erc20_transfers
2. •Filter by contract address in your pipeline transform:

transforms:
  usdc_only:
    type: sql
    primary_key: id
    sql: |
      SELECT * FROM source_name
      WHERE address = lower('0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913')

"I want all NFT activity on Ethereum"

Dataset: ethereum.erc721_transfers

"I want to monitor a specific smart contract"

1. •Dataset: <chain>.logs for raw events, or <chain>.decoded_logs for decoded events
2. •Filter by contract address in your transform

"I need multi-chain data"

Use multiple sources in your pipeline:

sources:
  eth_transfers:
    type: dataset
    dataset_name: ethereum.erc20_transfers
    version: 1.0.0
    start_at: latest
  base_transfers:
    type: dataset
    dataset_name: base.erc20_transfers
    version: 1.2.0
    start_at: latest

Troubleshooting

Dataset not found

Error: Source 'my_source' references unknown dataset 'invalid.dataset'

Fix:

1. •Check the chain prefix is correct (e.g., matic not polygon)
2. •Check the dataset type exists (e.g., erc20_transfers not erc20)
3. •Run goldsky dataset list to see all available options

Chain not listed

If you can't find a chain in the tables above:

goldsky dataset list | grep -i "<chain_name>"

Some chains use non-obvious prefixes (e.g., Polygon uses matic).

Version mismatch

Error: Version '2.0.0' not found for dataset 'base.erc20_transfers'

Fix: Check available versions:

goldsky dataset list | grep "base.erc20_transfers"

Use a version that exists in the output.

Related Skills

• •/turbo-pipelines - Create pipelines using discovered datasets
• •/goldsky-auth-setup - Set up CLI authentication first