API and RPC limitations

Selector extraction is heuristic #Link to Selector extraction is heuristic

Selectors are found by disassembling the runtime code and matching dispatcher comparisons. Unusual compilers, hand-written assembly or custom dispatch can hide functions. A verified function that extraction did not find is still listed, marked as not found in bytecode.

Explorer availability #Link to Explorer availability

Verification data comes from Blockscout, which is intermittently behind a Cloudflare bot challenge. ENTROPY does not bypass it. When Blockscout is unavailable it falls back to Sourcify; if neither answers, verification is shown as UNKNOWN with the reason.

Public RPC limits #Link to Public RPC limits

The public RPC is rate-limited. ENTROPY retries with backoff, staggers automatic reads, and samples the block height. Heavy use needs ENTROPY_RPC_URL. ENTROPY itself rate-limits clients and returns RATE_LIMITED with a retry time.

Latest-block reads #Link to Latest-block reads

All reads use the latest block at request time and say which block. There are no historical reads in this release.

In-memory caches #Link to In-memory caches

Results are cached in memory per server instance, keyed by address and code hash. Caches reset on restart and are not shared between instances.

Size limits #Link to Size limits

• Calldata: up to 128 KB.
• Logs per request: up to 500.
• Proxy hops followed: up to 3.
• Selectors looked up per contract: up to 512.

Not available yet #Link to Not available yet

Storage slot reads, disassembly listing, diamond facets, access-control findings, interface detection (ERC-20), deployer and balance, codegen targets other than viem, ABI compare, and transaction calldata decoding by hash.