Hosted indexing platform: deploy subgraphs (compatible with The Graph) and stream on-chain data into your own database via Mirror pipelines.
- 01hosted subgraphs without GRT
- 02migrating off Graph hosted service
- 03real-time data into Postgres / Snowflake / Kafka
- 04low-latency event streams
- 05long-tail and new EVM chains
- curl https://goldsky.com | sh
- pnpm add graphql graphql-request
| Variable | Scope | Description |
|---|---|---|
| GOLDSKY_API_KEY | Server | Goldsky API key (used by the CLI via `goldsky login` and for authenticated endpoints). |
Install the CLI with `curl https://goldsky.com | sh`, then `goldsky login`. Deploy a subgraph: `goldsky subgraph deploy <name>/<version> --path .`. To migrate from The Graph, use `goldsky subgraph deploy <name>/<version> --from-url <existing-query-url>`. For database streams, define a `pipeline.yaml` with sources (subgraph or direct chain) and sinks (Postgres, ClickHouse, Kafka, webhook) and run `goldsky pipeline apply pipeline.yaml`. Query subgraph endpoints with `graphql-request` against the URL printed by `goldsky subgraph list`.
- ⚑Goldsky subgraphs use the same manifest format as The Graph but the deploy URL and query URL are different — don't reuse Graph CLI deploy keys or gateway URLs.
- ⚑Mirror pipelines bill by rows processed and connector — backfilling a high-volume contract from genesis can produce a surprise bill; preview row counts first.
- ⚑Schema changes to a Mirror sink are not always auto-applied; some changes require dropping and re-syncing the destination table.
- ⚑Free tier limits the number of concurrent subgraphs and pipelines and pauses deployments after monthly quota; production should be on a paid plan.
- ⚑Chain coverage skews EVM — non-EVM chains (Solana, etc.) have limited or no support depending on the product surface.