Agent Tools

View .md

Risk Flags

The flags array in the response identifies specific risk conditions detected for the trade. Use flags to understand why a score is elevated and to make targeted decisions in your agent logic.

Flag Reference

Order Book Flags

Flag	Severity	Description
thin_book	Warning	Order book has insufficient depth relative to the trade size
high_slippage	Warning	Expected slippage is elevated
extreme_slippage	Critical	Expected slippage is very high, or trade exceeds available book depth

Liquidation Flags

Flag	Severity	Description
near_liquidation	Critical	Position is close to liquidation price
near_backstop	Critical	Position is close to the backstop liquidation threshold
high_leverage	Warning	Effective leverage is elevated
adl_risk	Warning	Auto-deleveraging risk detected for a profitable high-leverage position
partial_liquidation	Warning	Large position where partial liquidation would cause amplified slippage

Funding Flags

Flag	Severity	Description
funding_anomaly	Warning	Funding rate is statistically abnormal
funding_expensive	Warning	Projected annualized funding cost is elevated
cex_funding_divergence	Warning	Hyperliquid funding rate diverges significantly from CEX rates

Market Impact Flags

Flag	Severity	Description
high_impact	Warning	Expected market impact is elevated
low_volume	Warning	Trade size is large relative to the asset's recent volume

VaR Flags

Flag	Severity	Description
high_var	Warning	Value-at-Risk is elevated relative to notional
limited_history	Info	Insufficient historical data for full statistical analysis
new_listing	Info	Asset has limited trading history

Open Interest Flags

Flag	Severity	Description
oi_at_cap	Critical	Asset is at the exchange's open interest cap
oi_near_cap	Warning	Open interest is approaching the cap

Portfolio Flags

Flag	Severity	Description
concentration_risk	Warning	Portfolio is heavily concentrated in a single asset
correlated_exposure	Warning	Multiple positions are correlated in the same direction

Oracle Flags

Flag	Severity	Description
oracle_divergence	Warning	Mark price diverges significantly from oracle/index price

Data Quality Flags

Flag	Severity	Description
stale_cache	Info	Cached market data may be slightly delayed

Using Flags in Agent Logic

Flags let you build nuanced responses beyond just checking recommendation:

result = requests.post(...).json()

# Block on critical flags
critical = [f for f in result["flags"] if f in ["extreme_slippage", "near_liquidation", "near_backstop", "oi_at_cap"]]
if critical:
    agent.abort(f"Critical risk: {critical}")

# Adjust for warnings
if "thin_book" in result["flags"]:
    # Reduce size to match available liquidity
    size = min(size, result["slippage"]["depth_50bps"] * 0.5)

if "funding_expensive" in result["flags"]:
    # Only enter if expected alpha exceeds funding cost
    if expected_return < result["funding"]["annualized_pct"]:
        agent.skip("Funding cost exceeds expected return")

if "correlated_exposure" in result["flags"]:
    # Already exposed to this direction, reduce allocation
    size *= 0.5