If you like eRPC, give it a star on GitHub ⭐️
Use cases
Scale chains & providers
AI agents: fetch https://docs.erpc.cloud/use-cases/scale-chains-and-providers.llms.txt for the complete machine-readable version of this page (full configuration schema, defaults, worked examples, and source links). Append `.llms.txt` to any docs URL for the same treatment.AIFor agents: /use-cases/scale-chains-and-providers.llms.txt

Scale chains & providers

Supporting a new chain shouldn't be a project. Give eRPC one provider API key and it unlocks every chain that vendor supports — upstreams appear lazily the first time a request needs them. Behind the scenes, a scoring engine continuously ranks all your endpoints on real latency, errors, and how far behind the chain tip they are, so the next request always goes to whoever deserves it most. Auditioning a brand-new provider? Mirror live traffic to it in shadow mode before trusting it with a single real response.

Providers22 vendors built in — one key, all their chains.NetworksPer-chain settings and friendly aliases like /main/arbitrum.UpstreamsFine-grained control over each endpoint when you need it.Selection & scoringLive health metrics pick the best upstream, every time.Shadow upstreamsTest new providers on real traffic with zero user impact.

All of the above in one place — illustrative, not a tuned production config:

projects[]
erpc.yaml
projects:  - id: main    # one key → every chain the vendor supports, upstreams appear lazily    providers:      - endpoint: alchemy://${ALCHEMY_API_KEY}      - endpoint: drpc://${DRPC_API_KEY}    upstreams:      # mix in your own node alongside vendors      - id: my-own-node        endpoint: http://10.0.0.5:8545      # audition a new provider on 20% of live traffic, zero user impact      - id: candidate-provider        endpoint: https://rpc.new-vendor.xyz        shadow:          enabled: true          sampleRate: 0.2    networks:      - architecture: evm        evm: { chainId: 1 }        # /main/ethereum instead of /main/evm/1        alias: ethereum        # who deserves the next request, re-evaluated continuously        selectionPolicy:          evalFunc: |            (upstreams, ctx) => upstreams              .preferTag('!tier:fallback')              .sortByScore()

Agent reference

Copy one of these prompts into your AI agent session (Claude Code, Cursor, …) — each one points the agent at this page's machine-readable reference so it can do the work correctly:

Prompt Example #1: add new chains and providers from scratch
I want to add Alchemy and dRPC to my eRPC config so every chain they support is
available automatically, with live scoring picking the best upstream per request.
Work with my existing eRPC config. Read the full reference first:
https://docs.erpc.cloud/use-cases/scale-chains-and-providers.llms.txt
Prompt Example #2: shadow-test a new provider before trusting it
I want to audition a new RPC vendor on real traffic before giving it any production
requests. Configure a shadow upstream in my eRPC config that mirrors a sample of live
traffic to the candidate and logs mismatches, with zero user impact. Read the
reference first: https://docs.erpc.cloud/use-cases/scale-chains-and-providers.llms.txt
Prompt Example #3: tune scoring so fallback providers stay idle unless needed
I have primary and fallback providers in my eRPC config and I want the selection policy
to always prefer primaries and only route to fallbacks when primaries are degraded.
Walk through the scoring and tag config to make this work. Reference:
https://docs.erpc.cloud/use-cases/scale-chains-and-providers.llms.txt
Scale chains — agent starting pointsExpand for every option, default, and edge case — or copy this entire section into your AI assistant.

This page sells the outcome; implementation lives in the feature pages. Fetch their machine-readable companions:

  • Providers — every vendor's settings keys, URL shortcut schemes, lazy upstream generation.
  • Networks — network defaults inheritance, aliases, lazy bootstrap.
  • Upstreams — endpoint schemes, block availability gating, cordon reasons, defaults inheritance.
  • Selection & scoring — score components, multipliers, custom eval functions, tags.
  • Shadow upstreams — mirroring, sampling, mismatch observability.

Resolution chain an agent must know: provider shortcut endpoints expand to vendor-specific upstreams per network on first use; explicit upstreams and generated ones merge, with upstreamDefaults then upstream-level overrides applied; the selection policy orders the final set before every request. Full field tables live in each page's agent section.

Cut RPC cost & latencyTrust the data