onchain-analysis（链上分析）

SKILL: onchain-analysis

Purpose

Interpret blockchain data strategically: identify patterns, detect anomalies, map flows, and surface risk signals — data-backed only.

When to Use

• Wallet/contract behavior seems suspicious or unclear
• You need to understand fund flows before a decision
• Investigating market behavior, insider movement, or protocol health

Inputs

• wallet_data (optional): addresses + labels + balances
• contract_data (optional): contract address + ABI/artifacts + known roles
• transactions (required): tx list or tx ids/hashes
• chain (optional): chain + timeframe

Steps

1. Normalize inputs:

- ensure chain/time window is explicit - ensure transactions are uniquely identified

1. Identify patterns:

- recurring counterparties - periodic deposits/withdrawals - concentration and dispersion patterns

1. Detect anomalies:

- sudden large transfers - new counterparties with high volume - unusual contract interactions

1. Map flows:

- sources → sinks - intermediate hops - aggregator/bridge interactions (label as such)

1. Evaluate intent as hypotheses:

- propose 1–3 plausible explanations - attach confidence and what evidence would change it

1. Produce action-oriented output:

- risk signals - what to verify next

Validation

• Include tx hashes / block references when possible.
• Distinguish facts from hypotheses.
• If data is incomplete, state the missing pieces explicitly.

Output

• insights (facts + patterns)
• risk_signals
• opportunities (only if supported by data)
• hypotheses (with confidence)
• next_checks

Safety Rules

• Data-backed only; no “mind reading” claims.
• Do not assist illicit activity or evasion.

Example

Input: 200 txs for one wallet over 30 days. Output: “High concentration into 2 addresses; one new counterparty accounts for 70% volume; verify entity labels + bridge usage.”