← knowledge base
Risk & Safety·July 26, 2026·9 min read

Sybil attacks: how farming accounts are detected and analyzed

Risk & Safetyvideo © RDNE Stock project
Screen for sybils

Sybil attacks: how farming accounts are detected and analyzed

A sybil attack is the creation of many fake identities to game a system — airdrops, governance votes, reviews, reputation. On-chain, the evidence is in the shared structure: one controller, many masks.

The cluster signatures

  • Shared funding source — dozens of wallets all receiving first money from the same address or exchange withdrawal, often in the same block.
  • Coordinated timing — near-simultaneous activation, identical activity windows.
  • Mechanical behavior — identical transaction patterns, same protocols, same amounts.
  • Circular transactions — funds rotating through the cluster to simulate independent activity.
  • Uniform holdings — every wallet holding the same small amounts of the same tokens.

Detection approach

Detection is statistical, not accusatory:

  • Cluster by funding — group wallets by shared funding sources.
  • Measure independence — compare activity patterns within vs across clusters.
  • Score coordination — timing and amount overlap above a threshold is a strong signal.
  • Report confidence — a cluster with identical 0.5 SOL withdrawals from one address at the same second is high-confidence; loose correlation is not.

Why it matters to the ecosystem

Farming distorts airdrops, governance and analytics. Every metric a sybil inflates (volume, holders, engagement) is noise for everyone else. SIGBOT's cluster screen cleans the signal for genuine participants.

Your role

If you are evaluating an airdrop list, a trading community, or a partner's engagement, the same wallet-level screen applies: check funding clustering and coordination before trusting the numbers.

[Screen a list →](/analyze)

put it into practice
Analyze a real wallet

Airdrop farming, governance rigging, review manipulation — sybil networks manufacture identity at scale. Here is how cluster analysis exposes them.

Run a free analysis →
related constellation