Open-source multi-chain indexer with Optimism support

[tan-product]

Idea

Hey Optimism community,

We've been building custody and payment infrastructure at Fystack, and one problem kept coming up: reliable, real-time deposit detection across multiple chains without depending on third-party services or running separate indexing infrastructure per network.

So we built and open-sourced multichain-indexer — a single Go binary that indexes transactions across 12+ chains concurrently, including Optimism. It parses every block, matches destination addresses against your monitored set using bloom filters (sub-millisecond lookups against millions of addresses), and streams matched transfers to NATS JetStream for downstream processing.

It's built specifically for payment gateways, CEXs, and OTC desks that need production-grade deposit detection at scale. A few things that might be relevant to builders here:

• Optimism is supported and stable
• Four worker types handle real-time indexing, historical backfill, retries, and manual rescans
• RPC failover with exponential backoff built in
• Restart-safe: all progress persisted to KV store, picks up exactly where it left off
• Self-hosted, no external dependencies beyond your RPC endpoints
• We're at an early but stable stage and would love to get feedback from the Optimism builder community — especially around edge cases on OP specifically, integration patterns you'd expect, or anything missing for your use case.

If this solves a problem you've hit, give it a try and let us know how it goes. Contributors are very welcome.

👉 github.com/fystack/multichain-indexer

[noriki0202]

I really like the architecture here, especially the bloom-filter based address matching and the decision to keep everything as a single self-hosted binary instead of relying on external indexers.

I’ve been working around infra/automation systems recently, and the restart-safe worker design + RPC failover approach looks production-oriented compared to most deposit listeners I’ve seen.

Would be interested in contributing or experimenting with this further, particularly around high-throughput ingestion, mempool/pre-confirmation detection, or improving resiliency under unstable RPC conditions.
Perhaps are you planning support for stream replay/state snapshots or chain-specific optimizations for OP stack networks?