Code Snippet Reference

20 production-ready examples · TypeScript SDK · ψ → φ → α pattern

API Alignment

These snippets have been verified against the Memintel App Developer API v2.1. Methods marked [Python SDK] are available in the Python backend SDK only (POST /execute, POST /execute/batch, POST /execute/range, POST /execute/async, POST /evaluate/condition, POST /evaluate/condition/batch). The TypeScript SDK uses evaluateFull() as the primary execution path.

How to Read These Examples

Every snippet follows the same three-layer structure:

Layer	Role
ψ Concept	Computes a signal — what is measured and how
φ Condition	Interprets significance — what the signal means in context
α Action	Defines the automated response — what happens when φ fires

Each example shows: the SDK call, the response outputs, why the pattern works, and variations to adapt it.

A. Risk & Monitoring

1. Churn Risk Detection


ψ Concept	Compute churn probability score (0–1) for a user
φ Condition	Score > 0.8 over last 30 days = high churn risk
α Action	Send retention email when condition fires

import Memintel from "@memintel/sdk";

const client = new Memintel({ apiKey: process.env.MEMINTEL_API_KEY });

const result = await client.evaluateFull({
  concept_id: "org.churn_risk",
  concept_version: "1.2",
  condition_id: "org.high_churn",
  condition_version: "1.0",
  entity: "user_abc123",
  timestamp: "2024-03-15T09:00:00Z", // deterministic
});

result.result.value                              // 0.87 (churn score)
result.decision.value                            // true (condition fired)
result.decision.actions_triggered[0].status      // "triggered"
result.decision.actions_triggered[0].action_id   // "org.send_retention_email"

Why this works: Timestamp pins execution — the same call on any date returns the same result. Action fires automatically with no conditional branching in your code.

Variations:

Lower threshold to 0.65 → catch users earlier in churn trajectory
Use percentile condition instead → flag top 10% churners relative to cohort
Add dry_run: true → simulate without sending emails during testing

2. Fraud Risk Flagging


ψ Concept	Compute fraud probability score for a transaction
φ Condition	Score > 0.9 = high fraud risk
α Action	Freeze account and alert risk team

const result = await client.evaluateFull({
  concept_id: "org.fraud_score",
  concept_version: "3.1",
  condition_id: "org.fraud_above_threshold",
  condition_version: "2.0",
  entity: "txn_9f8e7d",
  timestamp: new Date().toISOString(),
  explain: true, // see which signals drove the score
});

const score = result.result.value;  // 0.94
const isFraud = result.decision.value;  // true
const actions = result.decision.actions_triggered;
// [{ action_id: "org.freeze_account", status: "triggered" },
//  { action_id: "org.alert_risk_team", status: "triggered" }]

// Inspect signal contributions
const contributions = result.result.explanation?.contributions;
// { velocity: 0.55, geo_mismatch: 0.28, device_risk: 0.17 }

Why this works: explain: true reveals which input signals drove the score. Two actions trigger in one atomic call — no orchestration code needed.

Variations:

Adjust threshold to 0.75 for higher recall / lower precision
Use z_score strategy on the condition to flag relative spikes vs. baseline
Use dry_run: true in staging to verify action payloads before going live

3. Credit Risk Monitoring


ψ Concept	Compute credit risk score for an account
φ Condition	Score crosses > 0.75 threshold (change strategy)
α Action	Trigger credit review workflow

// Evaluate the full pipeline — concept + condition + action
const result = await client.evaluateFull({
  concept_id: "org.credit_risk",
  concept_version: "1.0",
  condition_id: "org.credit_risk_elevated",
  condition_version: "1.0",
  entity: "account_456",
  timestamp: "2024-06-01T00:00:00Z",
});

if (result.decision.value) {
  console.log("Review triggered for:", result.decision.entity);
  console.log("Actions:", result.decision.actions_triggered);
  // [{ action_id: "org.trigger_credit_review", status: "triggered" }]
}

Python SDK

If the concept result is already cached, you can evaluate the condition alone using POST /evaluate/condition via the Python SDK — faster on repeated calls. Pre-warm with POST /execute to ensure the first condition evaluation hits cache.

Variations:

Use change strategy with direction: 'up' to only fire on deterioration
Batch-evaluate a portfolio via POST /evaluate/condition/batch (Python SDK)

4. Payment Failure Spike Detection


ψ Concept	Compute payment failure rate for a merchant over a rolling window
φ Condition	Z-score > 3.0 vs. historical baseline = anomalous spike
α Action	Page on-call team and open incident ticket

const result = await client.evaluateFull({
  concept_id: "org.payment_failure_rate",
  concept_version: "1.0",
  condition_id: "org.failure_rate_spike",
  condition_version: "1.0",
  entity: "merchant_abc",
  timestamp: "2024-06-15T14:00:00Z",
});

// condition uses z_score strategy — detects deviation from baseline
console.log(result.result.value);   // 0.23 (23% failure rate)
console.log(result.decision.value); // true (z-score = 3.8)
result.decision.actions_triggered[0].action_id  // "org.page_oncall"

Why this works: Z-score detects anomalies relative to baseline — not just absolute value. A 23% failure rate may be normal for some merchants; z-score cuts through that.

Variations:

Lower z-score threshold to 2.0 → more sensitive, higher alert rate
Change window from 1h to 15m → faster detection of fast-moving spikes
Add a second condition on raw rate > 0.15 for absolute floor protection

5. User Inactivity Alert


ψ Concept	Compute days since last meaningful engagement event
φ Condition	Days since activity > 14 = inactive
α Action	Send re-engagement push notification

const result = await client.evaluateFull({
  concept_id: "org.days_since_activity",
  concept_version: "1.0",
  condition_id: "org.user_inactive_14d",
  condition_version: "1.0",
  entity: "user_xyz",
  // no timestamp → snapshot mode, reflects current state
});

console.log(result.result.value);   // 17 (days since last session)
console.log(result.decision.value); // true
result.decision.actions_triggered[0].action_id  // "org.send_reengagement_push"

Why this works: Snapshot mode (no timestamp) reflects real-time state — ideal for operational monitoring where you want current data on every call.

Variations:

Segment by user tier: 7d for free users, 30d for paid users
Chain with a churn risk concept to suppress re-engagement if churn is already high

B. Business Metrics

6. Revenue Growth Tracking


ψ Concept	Compute revenue growth rate (month-over-month) for a company
φ Condition	Growth rate > 20% MoM = accelerating
α Action	Add company to high-growth outreach list

const result = await client.evaluateFull({
  concept_id: "org.revenue_growth_mom",
  concept_version: "1.0",
  condition_id: "org.high_growth_threshold",
  condition_version: "1.0",
  entity: "company_xyz",
  timestamp: "2024-03-31T23:59:59Z",
  explain: true,
});

result.result.value   // 0.34 (34% MoM growth)
result.decision.value // true (> 20% threshold)
result.decision.actions_triggered[0].action_id  // "org.add_to_outreach_list"

// Inspect signal contributions
const contributions = result.result.explanation?.contributions;
// { recurring_revenue: 0.68, new_logos: 0.22, expansion: 0.10 }

Python SDK

For concept-only execution with explain_mode: "summary", use POST /execute via the Python SDK. Both evaluateFull() and POST /execute share the same result cache — a concept executed by one is available to the other.

Variations:

Switch to a percentile condition to target the top 25% growers
Use POST /execute/range (Python SDK) to track growth trajectory over 12 months

7. Conversion Rate Drop


ψ Concept	Compute signup-to-paid conversion rate over a 7-day rolling window
φ Condition	Conversion rate drops > 20% vs. 30-day historical average (change strategy)
α Action	Notify product channel and create a tracking issue

const result = await client.evaluateFull({
  concept_id: "org.signup_conversion_rate",
  concept_version: "2.0",
  condition_id: "org.conversion_drop_alert",
  condition_version: "1.0",
  entity: "funnel_main",
  timestamp: "2024-07-10T00:00:00Z",
});

// change strategy — fires only on deterioration, not just low absolute value
console.log(result.result.value);   // 0.031 (3.1% conversion)
console.log(result.decision.value); // true (down from 4.2% baseline)
result.decision.actions_triggered[0].action_id  // "org.notify_product_channel"

Why this works: Change strategy is direction-aware — only fires on drops, not on a low absolute value that has been stable.

Variations:

Add direction: 'up' for a separate condition to detect conversion improvements
Segment by traffic source: entity = 'funnel_paid' vs 'funnel_organic'

8. Customer LTV Threshold


ψ Concept	Compute predicted lifetime value for a customer
φ Condition	Predicted LTV > $10,000
α Action	Assign dedicated success manager and upgrade support tier

// Evaluate LTV condition across a customer segment
// Note: evaluateConditionBatch() is available via the Python SDK.
// With the TypeScript SDK, evaluate each entity with evaluateFull().

const entities = ["cust_001", "cust_002", "cust_003", "cust_004"];

const decisions = await Promise.all(
  entities.map(entity =>
    client.evaluateFull({
      concept_id: "org.customer_ltv",
      concept_version: "1.0",
      condition_id: "org.high_value_customer",
      condition_version: "1.0",
      entity,
      timestamp: "2024-04-01T00:00:00Z",
    })
  )
);

const highValue = decisions.filter(d => d.decision.value === true);
console.log(highValue.map(d => d.decision.entity));
// ["cust_001", "cust_004"]
// Actions auto-triggered for matched entities

Python SDK

POST /evaluate/condition/batch (Python SDK) parallelises this natively in a single network call — more efficient for large segments.

Variations:

Use percentile condition to dynamically flag top 5% by LTV within a cohort
Scale to hundreds of entities using the Python SDK batch endpoint

9. Cohort Performance Comparison


ψ Concept	Compute 90-day retention rate per acquisition cohort
φ Condition	Retention rate in the bottom 25th percentile vs. all cohorts
α Action	Flag cohort for growth team review

const cohorts = ["cohort_jan24", "cohort_feb24", "cohort_mar24"];

// Step 1: Get concept values for all cohorts
// Note: executeBatch() is available in the Python SDK.
// With the TypeScript SDK, use parallel evaluateFull() calls.
const results = await Promise.all(
  cohorts.map(entity =>
    client.evaluateFull({
      concept_id: "org.cohort_90d_retention",
      concept_version: "1.0",
      condition_id: "org.low_retention_cohort",
      condition_version: "1.0",
      entity,
      timestamp: "2024-04-01T00:00:00Z",
    })
  )
);

results.forEach(r => {
  console.log(r.decision.entity, r.result.value, r.decision.value);
  // cohort_jan24  0.72  false
  // cohort_feb24  0.58  true  ← flagged
  // cohort_mar24  0.69  false
});

Why this works: Percentile condition is cross-sectional — ranks relative to peers, not an absolute cut.

10. KPI Anomaly Detection


ψ Concept	Compute daily active users (DAU) for the platform
φ Condition	DAU deviates > 2.5 standard deviations from 90-day rolling mean
α Action	Trigger executive alert and start anomaly investigation workflow

const result = await client.evaluateFull({
  concept_id: "org.daily_active_users",
  concept_version: "1.0",
  condition_id: "org.dau_anomaly_zscore",
  condition_version: "1.0",
  entity: "platform_main",
  timestamp: "2024-03-22T00:00:00Z",
  explain: true,
});

console.log(result.result.value);   // 84200 (DAU)
console.log(result.decision.value); // true (z-score = -3.1)

const contributions = result.result.explanation?.contributions;
// { mobile_sessions: 0.62, web_sessions: 0.38 }
result.decision.actions_triggered[0].action_id  // "org.exec_alert"

Why this works: Z-score adapts to seasonal variation — no manual baseline management. explain: true shows which channels drove the anomaly.

C. Time-Series & Trends

11. Moving Average Crossover


ψ Concept	Compute spread between 10-day and 50-day moving averages
φ Condition	Spread crosses from negative to positive (change strategy, direction: up)
α Action	Generate buy signal and notify strategy engine

const result = await client.evaluateFull({
  concept_id: "org.ma_spread_10_50",
  concept_version: "1.0",
  condition_id: "org.bullish_ma_crossover",
  condition_version: "1.0",
  entity: "AAPL",
  timestamp: "2024-03-15T16:00:00Z", // market close
});

// change strategy with direction: 'up' fires only on the crossover event
// — not on every day that 10d > 50d
console.log(result.result.value);   // 0.42 (spread: +$0.42)
console.log(result.decision.value); // true (just crossed zero)
result.decision.actions_triggered[0].action_id  // "org.generate_buy_signal"

Variations:

Reverse: direction: 'down' → detect bearish death cross
Use MACD spread instead of simple MA for momentum-adjusted signal
Evaluate across a basket of tickers using parallel evaluateFull() calls

12. Volatility Spike Detection


ψ Concept	Compute 10-day realised volatility (annualised) for an asset
φ Condition	Volatility > 2.5x the 90-day median (percentile strategy)
α Action	Alert risk desk and reduce position limits

const result = await client.evaluateFull({
  concept_id: "org.realised_volatility_10d",
  concept_version: "1.0",
  condition_id: "org.vol_spike_percentile",
  condition_version: "1.0",
  entity: "BTC-USD",
  timestamp: "2024-03-12T16:00:00Z",
});

console.log(result.result.value);   // 0.87 (87% annualised vol)
console.log(result.decision.value); // true (above 95th percentile)

const actions = result.decision.actions_triggered;
// [{ action_id: "org.alert_risk_desk", status: "triggered" },
//  { action_id: "org.reduce_position_limits", status: "triggered" }]

Why this works: Percentile condition adapts to the asset's own history — BTC at 87% vol is normal; SPX at 87% is extreme. Entity-relative comparison handles this automatically.

13. Momentum Acceleration


ψ Concept	Compute rate-of-change of 14-day RSI momentum
φ Condition	RSI momentum accelerating: week-over-week change > 15 points
α Action	Flag asset for entry watchlist review

// executeRange() for time-series history is available via the Python SDK.
// With the TypeScript SDK, use sequential evaluateFull() calls at each interval.

// Step 1: evaluate at weekly intervals to build the history
const dates = [
  "2024-02-01T00:00:00Z",
  "2024-02-08T00:00:00Z",
  "2024-02-15T00:00:00Z",
  "2024-03-01T00:00:00Z",
];

const history = await Promise.all(
  dates.map(timestamp =>
    client.evaluateFull({
      concept_id: "org.rsi_momentum_14d",
      concept_version: "1.0",
      condition_id: "org.momentum_accelerating",
      condition_version: "1.0",
      entity: "ETH-USD",
      timestamp,
    })
  )
);

history.forEach(r =>
  console.log(r.result.timestamp, r.result.value)
);
// 2024-02-01  42.1
// 2024-02-08  48.7
// 2024-02-15  56.3
// 2024-03-01  72.8  ← acceleration detected

// Step 2: evaluate the condition at the current decision point
const signal = await client.evaluateFull({
  concept_id: "org.rsi_momentum_14d",
  concept_version: "1.0",
  condition_id: "org.momentum_accelerating",
  condition_version: "1.0",
  entity: "ETH-USD",
  timestamp: "2024-03-15T16:00:00Z",
});

signal.decision.value  // true (acceleration confirmed)
signal.decision.actions_triggered[0].action_id  // "org.flag_for_watchlist"

Python SDK

POST /execute/range (Python SDK) handles the time-series loop natively and is more efficient for large date ranges.

14. Rolling Percentile Detection


ψ Concept	Compute 30-day cumulative return for a strategy
φ Condition	Return is in top 10th percentile vs. the strategy's own 1-year history
α Action	Promote strategy to primary allocation

const result = await client.evaluateFull({
  concept_id: "org.strategy_30d_return",
  concept_version: "1.0",
  condition_id: "org.top_decile_performer",
  condition_version: "1.0",
  entity: "strategy_momentum_v3",
  timestamp: "2024-03-31T23:59:59Z",
});

console.log(result.result.value);   // 0.187 (18.7% 30d return)
console.log(result.decision.value); // true (top decile)
result.decision.actions_triggered[0].action_id  // "org.promote_to_primary"

D. Trading & Finance

15. Stock Momentum Signal


ψ Concept	Compute composite momentum score combining price ROC and analyst sentiment
φ Condition	Composite score > 0.75 = strong momentum alignment
α Action	Generate trade signal and log to strategy ledger

const result = await client.evaluateFull({
  concept_id: "org.composite_momentum",
  concept_version: "2.0",
  condition_id: "org.strong_momentum_signal",
  condition_version: "1.0",
  entity: "NVDA",
  timestamp: "2024-03-15T16:00:00Z",
  explain: true,
});

const score = result.result.value;  // 0.81
const fired = result.decision.value; // true

const contrib = result.result.explanation?.contributions;
// { price_roc: 0.48, sentiment_score: 0.33 }
// When sentiment contribution > price: narrative is leading — early signal

Why this works: explain: true reveals whether price or narrative is leading. Composite concept fuses two distinct signals; condition acts on the combined output.

16. Narrative vs Price Divergence


ψ Concept	Compute divergence between sentiment score and normalised price momentum
φ Condition	Divergence > 0.3 with sentiment leading price
α Action	Alert research desk and flag for further analysis

const result = await client.evaluateFull({
  concept_id: "org.narrative_price_divergence",
  concept_version: "1.0",
  condition_id: "org.narrative_leading_price",
  condition_version: "1.0",
  entity: "TSLA",
  timestamp: "2024-03-10T16:00:00Z",
  explain: true,
});

const divergence = result.result.value;  // 0.42
const leading = result.decision.value;    // true

const breakdown = result.result.explanation?.contributions;
// { sentiment_component: 0.67, price_component: 0.25, vol_adj: 0.08 }
// High sentiment contribution: market hasn't caught up yet

Why this works: explain: true is essential here — the contribution breakdown IS the signal. When sentiment_component > price_component, the market hasn't priced in the narrative yet.

17. Sector Rotation Detection


ψ Concept	Compute relative strength spread between two sector ETFs (XLK vs. XLE)
φ Condition	Spread shifts by more than 0.15 over 5 days (change strategy)
α Action	Trigger sector rotation alert and rebalance workflow

// Execute concept for both sectors in parallel
const [tech, energy] = await Promise.all([
  client.evaluateFull({
    concept_id: "org.sector_momentum",
    concept_version: "1.0",
    condition_id: "org.sector_active",
    condition_version: "1.0",
    entity: "XLK",
    timestamp: "2024-03-15T16:00:00Z",
  }),
  client.evaluateFull({
    concept_id: "org.sector_momentum",
    concept_version: "1.0",
    condition_id: "org.sector_active",
    condition_version: "1.0",
    entity: "XLE",
    timestamp: "2024-03-15T16:00:00Z",
  }),
]);

console.log("Tech momentum:", tech.result.value);    // 0.72
console.log("Energy momentum:", energy.result.value); // 0.29

// Detect the rotation on the spread concept
const rotation = await client.evaluateFull({
  concept_id: "org.sector_spread_xlk_xle",
  concept_version: "1.0",
  condition_id: "org.sector_rotation_signal",
  condition_version: "1.0",
  entity: "sector_pair_xlk_xle",
  timestamp: "2024-03-15T16:00:00Z",
});

rotation.result.value   // 0.43 (spread, widening)
rotation.decision.value // true (rotation detected)

18. Event-Driven Trade Signal


ψ Concept	Compute post-earnings sentiment score for a ticker
φ Condition	Sentiment score > 0.65 within 2h of earnings release
α Action	Generate long signal and set position sizing parameters

// Earnings dropped at 16:00 — evaluate at 16:15 snapshot
const result = await client.evaluateFull({
  concept_id: "org.post_earnings_sentiment",
  concept_version: "1.0",
  condition_id: "org.strong_post_earnings_bullish",
  condition_version: "1.0",
  entity: "META",
  timestamp: "2024-02-01T21:15:00Z", // 15m after report
  explain: true,
});

// Deterministic: can re-evaluate this exact moment later for audit
console.log(result.result.value);   // 0.78 (post-earnings sentiment)
console.log(result.decision.value); // true (bullish signal)

const contrib = result.result.explanation?.contributions;
// { guidance_tone: 0.45, beat_magnitude: 0.32, mgmt_language: 0.23 }

E. Operations & Alerts

19. SLA Breach Detection


ψ Concept	Compute p99 API response latency over a 5-minute rolling window
φ Condition	p99 latency > 500ms = SLA breach
α Action	Page on-call engineer and open PagerDuty incident

const result = await client.evaluateFull({
  concept_id: "org.api_p99_latency_5m",
  concept_version: "1.0",
  condition_id: "org.sla_breach_500ms",
  condition_version: "1.0",
  entity: "service_payments_api",
  // no timestamp → snapshot mode for real-time ops monitoring
});

if (result.decision.value) {
  console.log("SLA BREACH:", result.result.value, "ms p99 latency");
  // SLA BREACH: 723 ms p99 latency
  console.log("Actions:", result.decision.actions_triggered);
  // [{ action_id: "org.page_oncall", status: "triggered" },
  //  { action_id: "org.open_pagerduty_incident", status: "triggered" }]
}

Why this works: Snapshot mode (no timestamp) ensures real-time monitoring reflects current state. Two actions trigger atomically — page + incident creation with no coordination code.

Variations:

Add a warning condition at 300ms for early alerting before SLA breach
Segment by endpoint: entity = 'service_payments_api/checkout'

20. System Anomaly Detection


ψ Concept	Compute composite anomaly score from CPU, memory, and error rate signals
φ Condition	Composite z-score > 3.0 = system anomaly
α Action	Trigger auto-remediation workflow and create incident

// For lightweight concepts, evaluateFull() is the right default.
// For compute-heavy multi-signal concepts that may exceed 30s,
// use POST /execute/async + GET /jobs/{job_id} via the Python SDK.

const result = await client.evaluateFull({
  concept_id: "org.system_anomaly_composite",
  concept_version: "1.0",
  condition_id: "org.system_anomaly_detected",
  condition_version: "1.0",
  entity: "cluster_prod_us_east_1",
  // no timestamp → snapshot mode for real-time ops
});

console.log(result.result.value);   // 0.91 (composite anomaly score)
console.log(result.decision.value); // true (z-score = 3.7)
result.decision.actions_triggered[0].action_id  // "org.trigger_auto_remediation"
result.decision.actions_triggered[1].action_id  // "org.create_incident"

Python SDK — Async Execution

For heavy multi-signal concepts expected to exceed 30 seconds, use POST /execute/async + GET /jobs/{job_id} via the Python SDK. Poll until status === "completed", then evaluate the condition using POST /evaluate/condition — it will hit the cached result from the async job. Cancel a running job with DELETE /jobs/{job_id}.

SDK Method Quick Reference

Method	Description
`client.evaluateFull()`	ψ + φ + α in one atomic call — recommended default (TypeScript SDK)
`POST /execute`	ψ only, with optional `explain` — Python SDK
`POST /evaluate/condition`	φ only, uses cached ψ result — Python SDK
`POST /evaluate/condition/batch`	φ for multiple entities in parallel — Python SDK
`POST /execute/batch`	ψ for multiple entities in parallel — Python SDK
`POST /execute/range`	ψ over a time range (time-series) — Python SDK
`POST /execute/async`	ψ as a background job (> 30s workloads) — Python SDK
`GET /jobs/{job_id}`	Poll an async job for result — Python SDK

Base URL: https://api.memsdl.ai/v1 · Auth: X-API-Key header · Install: npm install @memintel/sdk

How to Read These Examples​

Categories​

A. Risk & Monitoring​

1. Churn Risk Detection​

2. Fraud Risk Flagging​

3. Credit Risk Monitoring​

4. Payment Failure Spike Detection​

5. User Inactivity Alert​

B. Business Metrics​

6. Revenue Growth Tracking​

7. Conversion Rate Drop​

8. Customer LTV Threshold​

9. Cohort Performance Comparison​

10. KPI Anomaly Detection​

C. Time-Series & Trends​

11. Moving Average Crossover​

12. Volatility Spike Detection​

13. Momentum Acceleration​

14. Rolling Percentile Detection​

D. Trading & Finance​

15. Stock Momentum Signal​

16. Narrative vs Price Divergence​

17. Sector Rotation Detection​

18. Event-Driven Trade Signal​

E. Operations & Alerts​

19. SLA Breach Detection​

20. System Anomaly Detection​

SDK Method Quick Reference​

How to Read These Examples

Categories

A. Risk & Monitoring

1. Churn Risk Detection

2. Fraud Risk Flagging

3. Credit Risk Monitoring

4. Payment Failure Spike Detection

5. User Inactivity Alert

B. Business Metrics

6. Revenue Growth Tracking

7. Conversion Rate Drop

8. Customer LTV Threshold

9. Cohort Performance Comparison

10. KPI Anomaly Detection

C. Time-Series & Trends

11. Moving Average Crossover

12. Volatility Spike Detection

13. Momentum Acceleration

14. Rolling Percentile Detection

D. Trading & Finance

15. Stock Momentum Signal

16. Narrative vs Price Divergence

17. Sector Rotation Detection

18. Event-Driven Trade Signal

E. Operations & Alerts

19. SLA Breach Detection

20. System Anomaly Detection

SDK Method Quick Reference