[
  {"query": "what's the full YAML syntax for a postgres sink? what fields are required vs optional?", "should_trigger": true},
  {"query": "what does the checkpoint field do in a turbo pipeline config?", "should_trigger": true},
  {"query": "what are all the top-level fields in a turbo pipeline YAML file?", "should_trigger": true},
  {"query": "how do i configure a clickhouse sink with the order_by field? what's the syntax?", "should_trigger": true},
  {"query": "getting 'unknown field: primary_keys' validation error, what's the correct field name?", "should_trigger": true},
  {"query": "what's the difference between start_at: earliest vs start_at: latest in a dataset source?", "should_trigger": true},
  {"query": "how do i set up a kafka sink? what fields does it need — topic, partitions, etc?", "should_trigger": true},
  {"query": "what's the correct way to reference a transform output in a sink's `from:` field?", "should_trigger": true},
  {"query": "should i use a kafka source or a dataset source for tracking live token balances on ethereum?", "should_trigger": true},
  {"query": "i need to fan out from one source into 3 different sinks — postgres for queries, kafka for streaming, and s3 for archival. how do i design that?", "should_trigger": true},
  {"query": "should i use one pipeline for all 5 chains or separate pipelines per chain? what are the tradeoffs?", "should_trigger": true},
  {"query": "what resource size should i use for a pipeline that decodes 10 different event types with UNION ALL?", "should_trigger": true},
  {"query": "i'm not sure whether to use streaming mode or job mode for my use case — i need a full historical backfill plus continued live updates", "should_trigger": true},
  {"query": "help me think through the architecture for a multi-chain NFT tracker that feeds both a postgres API database and a clickhouse analytics warehouse", "should_trigger": true},
  {"query": "what's the best sink for real-time aggregate balances that need to update incrementally?", "should_trigger": true},
  {"query": "what are all the available sink types i can write to?", "should_trigger": true},
  {"query": "how do i set up a mysql sink? do i need a primary_key for upserts?", "should_trigger": true},
  {"query": "what fields does the sqs_sink take and how do i point it at a queue?", "should_trigger": true},
  {"query": "walk me through building and deploying a complete pipeline for base erc20 transfers", "should_trigger": false},
  {"query": "how do i write a SQL transform to decode EVM event logs?", "should_trigger": false},
  {"query": "my pipeline is throwing schema mismatch errors, help me debug and fix it", "should_trigger": false},
  {"query": "what's the dataset name for polygon raw transactions?", "should_trigger": false},
  {"query": "how do i write a typescript transform to filter records by amount?", "should_trigger": false},
  {"query": "how do i create a secret for my postgres connection?", "should_trigger": false},
  {"query": "ok i've decided on the architecture, now help me actually build and deploy the pipeline YAML", "should_trigger": false},
  {"query": "my pipeline is in error state, help me figure out why it crashed", "should_trigger": false},
  {"query": "walk me through deploying this pipeline to goldsky", "should_trigger": false},
  {"query": "when should i use a dynamic table vs a regular sql transform join?", "should_trigger": false}
]