FORENSIAWeb3 forensicsOpen workspace

Budgets and bounded lookups

Every lookup has a fixed allowance of requests, time and data. Here are the numbers and what happens when one runs out.

3 min read · How a lookup works

#Plain languageIn plain terms

A busy address can have millions of records. No free tool reads all of them. Forensia reads a bounded slice and is explicit about where the slice ends.

Each lookup gets a fixed allowance. When it is spent the lookup stops and says so. It does not quietly return a partial answer dressed up as a full one.

#How it worksHow it works

A counter travels with each lookup. Every outgoing request, including a network probe, a fallback attempt and a repeated attempt at the Bitcoin source, spends one. Steps that are essential stop the lookup when the counter is empty. Optional steps are skipped and the result notes what was left out.

#Technical detailThe exact rules

Kind of lookupRequests allowed
Standard lookup, public deployment48
Standard lookup, other deployments128
Follow native value onward32
Follow a token onward24
Across bridge check48
Bitcoin branch step2

Checking other sources after one returns no record for a transaction hash can spend more requests on a hash that is not there. The allowance is not raised to make room for this.

A lookup runs for at most 30 seconds. A server process handles at most 4 lookups at once, or 3 on the public deployment.

How far back an address lookup reads

NetworkRecent full blocksToken event window
Ethereum1812,000 blocks
Base, Arbitrum, Optimism2415,000 blocks
Polygon2412,000 blocks
BNB Smart Chain, Avalanche188,000 blocks

If a provider caps how many blocks one log query may span, the windows shrink to fit and the result says so. On the public deployment recent full blocks are not scanned at all; native activity comes from indexed transfer pages.

When no transfer index supplied history for an EVM address, Overview states the reach under "How far back this looked": how many of the newest blocks were read for native transfers, or that neither an index nor a recent-block scan was available so native transfers were not looked for.

Other ceilings

WhatLimit
EVM address records retained32
Activity / receipt rows per page12
Transaction steps per page20
Transactions checked against a receipt4, or 2 on the public deployment
Call trace depth6
Call frames kept64
Token-event items kept per transaction2,048; up to 256 per batch log
Token contracts asked for their name4
Approval events kept48

Public beta limits

LimitValue
Completed lookups per day, everyone750
Attempts per hour, per client8
Completed lookups per day, per client30
Lookups running at once, per client1
Outgoing requests per day, everyone15,000

The full 48 requests are reserved when a lookup starts and the unused part is refunded when it ends. A client is identified by a keyed hash of its network address, never by the address itself.

#LimitsWhat it does not tell you

  • When the hard request budget is hit, the lookup ends with a clear error and can be retried. "Retry lookup" repeats the same identifier and network choice as a new lookup, which gets its own allowance and counts toward the public beta limits again.
  • When an optional step is skipped, the result still completes. Read its coverage notes to see what was skipped.
  • The 32 rows shown are a balanced sample: up to 12 transfers, 12 token events and 8 contract events, newest first.