Circuit

Liquidation-defense circuit breaker for Solana credit positions.
Circuit watches health factors, scores risk in real time, and turns a drifting position into a concrete action plan before liquidation becomes a race.

Liquidations do not start at the liquidation line. They start when a position stops being comfortably safe and nobody has a system that treats that drift as operationally urgent.

Circuit exists to make that slide impossible to ignore. It monitors Solana credit positions, maps them into explicit defense states, and tells the operator what the correct posture is before the position turns into a scramble.

FETCH -> ASSESS -> CLASSIFY -> MATCH RULES -> RESPOND

Risk Dashboard • Circuit Breaker Rules • Why Circuit Exists • Defense Posture • At a Glance • Escalation Ladder • How Operators Use Circuit • Example Output • Rule Design • Risk Controls • Quick Start

Risk Dashboard

Circuit Dashboard

Circuit Breaker Rules

Circuit Rules

Why Circuit Exists

Most DeFi monitoring stacks are observational. They tell you what happened to the position. Circuit is built for the next question: what should happen now?

That sounds simple until the wallet is real. One lending leg weakens faster than expected, another position is still technically fine, and the operator now has to decide whether to watch, reduce, close, or add collateral. If that logic lives only in someone's head, it usually fails under stress.

Circuit moves that logic into a standing system. The point is not to maximize leverage. The point is to keep leveraged or borrowed exposure from drifting into forced behavior.

Defense Posture

Circuit should feel less like a dashboard and more like a control layer.

The product is useful when it does three things well:

turns position drift into a named risk state
maps each risk state to a clear response
separates cautionary alerts from situations where delay is actually dangerous

That is why the repo is framed as a circuit breaker. A breaker is not there to predict every problem. It is there to stop a manageable problem from becoming a catastrophic one.

At a Glance

Use case: liquidation defense for leveraged or borrowed Solana positions
Primary input: collateral, debt, health factor, liquidation threshold, and protocol-specific position state
Primary failure mode: watching health drift too late and reacting after liquidation risk is already urgent
Best for: operators managing lending exposure on Kamino and MarginFi who want structured escalation

The Escalation Ladder

Level	Health Factor	What it means operationally	Typical next move
`SAFE`	above watch threshold	no active problem yet	monitor only
`WATCH`	drifting weaker	situation deserves attention	review but do not overreact
`WARNING`	deterioration is material	delay is becoming expensive	reduce debt or add collateral
`CRITICAL`	liquidation risk is near	the position needs action now	close, repay, or auto-exit

The important thing is not the label itself. It is the fact that each label changes behavior.

How Operators Use Circuit

As A Live Defense Layer

This is the most obvious use. The wallet is running real leverage, the operator wants early warnings, and the board needs to say when the correct answer changed from "watch this" to "do something now."

As A Policy Check

Circuit is also useful before auto-exit is enabled. Teams often want the system to print the right action for a while before they trust it to take the action.

As A Credit Discipline Tool

Even when no one is about to get liquidated, Circuit forces a useful question: are these thresholds and these position sizes still acceptable for the way the operator actually trades?

How It Works

Circuit runs a defense loop every cycle:

pull the latest supported lending positions for the monitored wallet
compute health, liquidation distance, and position-level risk
compare the current state against the configured breaker thresholds
summarize the position as an operational situation, not just a number
print or execute the response that matches the actual danger level

The system should be judged on whether it creates the right response hierarchy. A good credit monitor is not one that shouts the loudest. It is the one that reacts correctly as the position weakens.

Before You Enable Auto-Exit

Circuit does not need to be fully automated on day one to be valuable. In fact, most operators should start with alert-only mode and check whether the ladder matches how they already think about risk.

Ask four questions before enabling automatic execution:

are the health thresholds actually correct for this portfolio
does the printed action match what the operator would do manually
are the scanned venues the real sources of liquidation risk
is the wallet comfortable with speed over discretion in critical states

If those answers are weak, the right first step is still using Circuit as a standing advisor.

What A Strong Deployment Looks Like

positions are checked often enough that warning states are actually useful
the watch and warning levels are far enough apart to change behavior
the operator sees exact positions and exact next actions, not generic fear language
auto-exit is reserved for the states where hesitation is more dangerous than overreaction

That is what makes the product believable to non-dev readers too. The value is obvious because the consequence is obvious.

Example Output

CIRCUIT // POSITION ALERT

protocol           kamino
position           SOL borrow against JTO collateral
health factor      1.07
risk level         warning
distance to liq    6.1%

recommended action:
- repay part of the debt leg
- or add fresh collateral within the next cycle

Rule Design

Circuit is more useful when the rule design is explicit.

watch threshold should catch deterioration early enough to matter
warning threshold should represent a state where delay now has cost
critical threshold should mean the position is no longer safe to leave unattended
auto-exit should be reserved for operators who already trust the policy layer

The point of rule design is not elegance. It is operational clarity under pressure.

What Circuit Refuses To Pretend

it does not claim to cover protocols it is not actually scanning
it does not call a position safe merely because it is not critical yet
it does not confuse generic notifications with liquidation defense
it does not assume the same health thresholds fit every wallet

That honesty helps the repo read more like a product than a pitch deck.

Risk Controls

threshold-based escalation: each risk class is tied to explicit health cutoffs
auto-exit off by default: operators can start in alert-only mode before enabling execution
position-level recommendations: recommendations are attached to the actual position instead of a generic market warning
continuous polling: the system is biased toward early detection rather than after-the-fact reporting

Circuit is best used as a standing control layer. If a position is already deep in the danger zone when monitoring begins, the system can still help, but it is no longer solving the easy part.

Quick Start

git clone https://github.com/CircuitSOL/Circuit
cd Circuit
bun install
cp .env.example .env
bun run dev

Configuration

ANTHROPIC_API_KEY=sk-ant-...
WALLET_ADDRESS=your-wallet
CHECK_INTERVAL_MS=15000
WATCH_HEALTH_FACTOR_THRESHOLD=1.2
WARNING_HEALTH_FACTOR_THRESHOLD=1.1
CRITICAL_HEALTH_FACTOR_THRESHOLD=1.05
AUTO_EXIT=false
LOG_LEVEL=info

Support Docs

License

MIT

the best liquidation defense starts before the position looks dramatic.