SkyInsights Risk Intelligence

Use {skillDir}/scripts/skyinsights.py for all API calls. The script handles credentials, request formatting, error handling, screening polling, and terminal-friendly output.

Running Commands

python {skillDir}/scripts/skyinsights.py <subcommand> [args...]

Subcommands

• kya <address> [chain=eth] — GET /v4/kya/risk
• labels <address> [chain=eth] — GET /v4/kya/labels
• screen <address> [chain=eth] [rule_set_id=standard-mode-rule-set] — GET /v4/kya/screening_v2 + poll (⚠️ 5–15s)
• kyt <txn_hash> <chain> — GET /v4/kyt/risk
• help — show usage

rule_set_id options: standard-mode-rule-set (default), fast-mode-rule-set

Supported Chains

Per-endpoint chain support. Use the API Value when specifying the chain parameter.

kya/screening_v2 supports: btc eth polygon op arb avax bsc tron wemix base multi-chain is supported by kya/risk only.

Default Workflow

Route user requests as follows:

• User mentions a transaction hash / tx / txn → run kyt
• User asks about labels / entity / institution / exchange for an address → run labels
• User asks if an address is risky / safe / suspicious (no subcommand specified):
  1. Run kya first — fast, checks the address's own labels and risk score.
  2. If kya returns None or Low, automatically follow with screen — a clean address can still have counterparty exposure to sanctioned or hacked funds. Tell the user: "该地址自身风险较低，正在进行合规筛查以检查对手方暴露情况，请稍候（约 5–15 秒）…"
  3. If kya returns Medium, High, or Unknown, present the result directly without running screen.

Output Expectations

Printed shapes per subcommand:

# kya
Risk: {None|Low|Medium|High|Unknown} (score={0-5})
Reasons: {reason1}, {reason2}     ← omitted if empty

# labels
Entity: {name}  Type: {type}
Labels: {label1}, {label2}        ← omitted if none

# screen
Screening: {Pass|Fail|Pending}
Flagged counterparties: {count}
{counterparty_address}  {risk_level}  {reason}  ← one line per flagged party

# kyt
Tx Risk: {None|Low|Medium|High|Unknown} (score={0-5})
Transfers: {count}
{from} → {to}  {amount} {token}   ← one line per transfer

Risk level emojis: ✅ None / 🟡 Low / 🟠 Medium / 🔴 High / ⚪ Unknown

After running the script, summarize the key findings in plain language — especially the risk level and notable risk factors. Don't just repeat the raw output.

Error Handling

• Missing credentials: tell the user to set SKYINSIGHTS_API_KEY and SKYINSIGHTS_API_SECRET.
• Unsupported chain: ask the user to retry with one of the supported chain identifiers.
• risk_level is Unknown: the address or transaction is not indexed on that chain, or has no available risk history.

Risk Factors Reference

Risk Score & Level Mapping

Risk Reasons Format

The risk_reasons field explains why a risk was assigned. Each entry is prefixed by type:

• label: Scam/Rugpull — risk comes from a specific label
• label: Sanction/OFAC
• entity: huione — risk linked to a known entity without a direct label
• entity: blender_io

Usage (for users)

When a user asks how to use this skill, show them these commands:

/skyinsights kya <address> [chain]       Address risk assessment / 地址风险评分
/skyinsights labels <address> [chain]    Address labels & entity info / 地址标签和实体信息
/skyinsights screen <address> [chain]    Compliance screening (AML) / 合规筛查，需 5–15 秒
/skyinsights kyt <txn_hash> <chain>      Transaction risk / 交易风险分析
/skyinsights help                        Show help / 显示帮助

Prerequisites / 前提条件： Set credentials in environment or .env file / 在环境变量或 .env 文件中设置凭证：

SKYINSIGHTS_API_KEY=your_key
SKYINSIGHTS_API_SECRET=your_secret