Skip to main content

Squillify development is paused indefinitely. Read why, and discover what this experiment achieved. Read the story

MicroSkills Reference

Error Budget Burn Rate

observability.evaluate_slo_burn

ProductionBounded custom (Class B)Deterministicstrictinstantv1.0.0
No network

Overview

You send a series, an objective and the instant the windows end, and we return the burn rate, an alert level and every number the decision used. A level fires only when every window configured for it is at or above its threshold, which is the whole point of multi-window alerting: a single short window at 14.4 is a blip, and paging on it is how an alert becomes noise that gets muted. A window whose data covers less than your minimum says so instead of extrapolating, so a five-minute deploy failure never becomes a one-hour burn rate. There is no clock here; every window measures back from the instant you send.

Burn-rate alerting is easy to get subtly wrong: the multi-window rule is an AND and gets implemented as an OR, a spent budget is a state and gets treated as a rate, and a short series gets extrapolated into a page.

Reliability

Fully deterministic and clock-free: the same series and windowEnd always give the same level, and re-running tomorrow reproduces today's answer exactly. Levels are evaluated in the precedence BUDGET_EXHAUSTED over PAGE over TICKET over NONE. Failures, none worth retrying unchanged: OBS_SLO_SERIES_INVALID means series was absent or empty, a point carried neither shape or both, good exceeded total, or an error ratio fell outside 0 to 1. OBS_SLO_SERIES_SHAPE_MIXED means the series mixed counted points with ratio points, and the answer names the index where it changed. OBS_SLO_SERIES_LIMIT_EXCEEDED means more than 10000 points arrived, so pre-aggregate into coarser buckets. OBS_SLO_TIMESTAMP_INVALID means windowEnd or a point timestamp was absent, unparseable or carried no offset. OBS_SLO_OFFSET_MIXED means the series mixed offsets under REQUIRE_UNIFORM or carried a non-UTC one under REQUIRE_UTC, with both positions named. OBS_SLO_OBJECTIVE_INVALID means the objective was absent, outside 0.5 to 0.999999, or carried more than 6 decimal places. OBS_SLO_PERIOD_INVALID means periodDays was absent or outside 1 to 366. OBS_SLO_WINDOW_INVALID means a window entry lacked a name, duration or level, or repeated a name. OBS_SLO_WINDOW_LIMIT_EXCEEDED means more than 4 windows arrived, which is the published cap. OBS_SLO_THRESHOLD_INVALID means a burn threshold fell outside 0.1 to 1000, or minCoverageRatio outside 0 to 1. OBS_SLO_SAMPLE_NON_FINITE means an error ratio was a not-a-number or an infinity, and the answer names its index. OBS_SLO_OPTION_UNSUPPORTED means a property or an offset policy is not one we accept.

Context footprint

The answer is one level, one budget block and at most four window rows, so it stays the same size whatever the series length. A successful call costs 1 execution credit.

When to use it

When you hold a good-and-total or error-ratio series and want a defensible alert level with every input to the decision visible.

When you still need to compute the indicator, or when you want a percentile rather than a ratio.

Scope

Note
Compute error-budget burn over up to four caller-declared windows and return one alert level with the numbers behind it.

Non-goals

  • Does not fetch a metric, send an alert or open a ticket
  • Does not read a clock — the windows end where you say they end
  • Does not extrapolate a short window into a longer one
  • Does not decide what caused the burn or what to do about it
  • Does not accept an objective of 1, which would make every rate infinite

Properties

IdempotentRetry safeStreaming supportedFile input supportedRequires external service

Capabilities

analysis

Cost & billing

Flat rate per invocation

MetercallsBillable onsuccessCost per call1 credit

File handling

Accepts file_refProduces file_refMax input1 MB

How we handle your data

Sensitive output possible
What this means

This skill may receive sensitive content and may return submitted, transformed, extracted, or reformatted content.

Squillify guarantees
  • Runs inside Squillify
  • This skill runs inside Squillify.
  • Payloads are not sent to an external service by this skill.
  • Squillify does not permanently store submitted payloads.
  • Telemetry records metadata only, not submitted content.
  • Submitted payloads are not used for training.
  • Submitted payloads are not reviewed by humans.
  • This skill does not target sensitive data.
  • The response may contain submitted or transformed content.
Your responsibility

Squillify does not determine whether submitted content is sensitive. Your application is responsible for deciding what data it is allowed to submit. Your application is responsible for securing the response.

Read the Microskill Data Handling Policy

Input schema

FieldTypeRequiredDescription
seriesarrayYesSeries points, 1 to 10000 closed objects each carrying t plus either good and total or errorRatio.
sloobjectYesClosed SLO: objective between 0.5 and 0.999999 and periodDays between 1 and 366.
windowEndstringYesISO-8601 instant every window measures back from.
windowsarrayNoUp to 4 evaluation windows; omit it for the published default fast, fast_confirm, medium and slow set.
thresholdsobjectNoClosed default burn thresholds; omit it for pageBurnRate 14.4 and ticketBurnRate 1.
minCoverageRationumberNoShare of a window its data must span before it may trip; omit it for 0.5.
scaleintegerNoDecimal places from 0 through 12; omit it for 6.
offsetPolicystringNoOffset mixing rule; omit it for REQUIRE_UNIFORM.

Output schema

FieldTypeRequiredDescription
burnRatenumberYesRate of the highest-precedence tripping window, or the shortest window when none trips, or JSON null when no window held a point.
alertLevelstringYesNONE, TICKET, PAGE or BUDGET_EXHAUSTED in rising precedence.
dataSufficiencystringYesSUFFICIENT, PARTIAL or INSUFFICIENT.
errorBudgetobjectYesClosed budget block with objective, periodDays, budgetRatio, budgetEvents, consumedBudgetRatio and remainingBudgetRatio.
windowResultsarrayYesOne closed result per configured window in configuration order, 1 to 4 entries.
summaryobjectYesClosed receipt with windowEnd as parsed, seriesPointCount, inWindowPointCount, scale and roundingMode.

API contract

POST /v1/execute
curl
curl -sS -X POST "$BASE/v1/execute" -H "Authorization: Bearer $KEY" -H 'Content-Type: application/json' -d '{"skillKey":"observability.evaluate_slo_burn","input":{"series":[{"t":"2026-07-29T09:30:00Z","good":990,"total":1000}],"slo":{"objective":0.99,"periodDays":30},"windowEnd":"2026-07-29T10:00:00Z"}}'
Integrations
TypeScript SDK
import { Squillify } from "@squillify/sdk";

const sq = new Squillify({ apiKey: process.env.SQUILLIFY_API_KEY });

const { output } = await sq.execute("observability.evaluate_slo_burn", {"slo":{"periodDays":30,"objective":0.99},"series":[{"total":1000,"good":900,"t":"2026-07-29T09:05:00Z"},{"total":1000,"good":900,"t":"2026-07-29T09:57:00Z"},{"total":1000,"good":900,"t":"2026-07-29T09:59:00Z"}],"minCoverageRatio":0,"windowEnd":"2026-07-29T10:00:00Z"});

Errors

CodeHTTPRetryableDescription
VALIDATION_ERROR400NoThe request payload failed schema validation.
UNAUTHORIZED401NoMissing or invalid API key.
SKILL_NOT_ACCESSIBLE403NoThe API key is not granted access to this skill.
SKILL_NOT_AVAILABLE_FOR_PLAN403NoThis skill is not included in your current plan.
PLAN_ELIGIBILITY_UNAVAILABLE503YesPlan eligibility could not be decided. Retry after backoff.
SKILL_NOT_FOUND404NoThe requested skill or version does not exist.
TENANT_RATE_LIMITED429YesRequest rate limit exceeded. Retry after backoff.
QUOTA_EXCEEDED429YesPlan quota exhausted. Retryable once the quota period resets.
EXECUTION_TIMEOUT504YesExecution exceeded the allowed time budget.
OBS_SLO_SERIES_INVALID400NoThe request payload failed this skill's input validation.
OBS_SLO_SERIES_SHAPE_MIXED400NoThe request payload failed this skill's input validation.
OBS_SLO_SERIES_LIMIT_EXCEEDED429NoA size, quota, rate, or time limit was exceeded.
OBS_SLO_TIMESTAMP_INVALID400NoThe request payload failed this skill's input validation.
OBS_SLO_OFFSET_MIXED400NoThe request payload failed this skill's input validation.
OBS_SLO_OBJECTIVE_INVALID400NoThe request payload failed this skill's input validation.
OBS_SLO_PERIOD_INVALID400NoThe request payload failed this skill's input validation.
OBS_SLO_WINDOW_INVALID400NoThe request payload failed this skill's input validation.
OBS_SLO_WINDOW_LIMIT_EXCEEDED429NoA size, quota, rate, or time limit was exceeded.
OBS_SLO_THRESHOLD_INVALID400NoThe request payload failed this skill's input validation.
OBS_SLO_SAMPLE_NON_FINITE400NoThe request payload failed this skill's input validation.
OBS_SLO_OPTION_UNSUPPORTED400NoThe skill does not support the requested variant.

Other limits

LimitValue
Recommended inline bytes32 KiB

Limitations

  • Does not fetch a metric, send an alert or open a ticket
  • Does not read a clock — the windows end where you say they end
  • Does not extrapolate a short window into a longer one
  • Does not decide what caused the burn or what to do about it
  • Does not accept an objective of 1, which would make every rate infinite
  • Computing the indicator itself, which observability.calculate_sli answers
  • Summarising raw latency samples, which observability.summarize_metrics_window answers

How to call

POST /v1/execute (requires authentication)

Request bodyjson
{
  "skillKey": "observability.evaluate_slo_burn",
  "input": {
    "slo": {
      "periodDays": 30,
      "objective": 0.99
    },
    "series": [
      {
        "total": 1000,
        "good": 900,
        "t": "2026-07-29T09:05:00Z"
      },
      {
        "total": 1000,
        "good": 900,
        "t": "2026-07-29T09:57:00Z"
      },
      {
        "total": 1000,
        "good": 900,
        "t": "2026-07-29T09:59:00Z"
      }
    ],
    "minCoverageRatio": 0,
    "windowEnd": "2026-07-29T10:00:00Z"
  }
}

Examples

A burn that clears the ticket thresholds but neither page threshold

Inputjson
{
  "slo": {
    "periodDays": 30,
    "objective": 0.99
  },
  "series": [
    {
      "total": 1000,
      "good": 900,
      "t": "2026-07-29T09:05:00Z"
    },
    {
      "total": 1000,
      "good": 900,
      "t": "2026-07-29T09:57:00Z"
    },
    {
      "total": 1000,
      "good": 900,
      "t": "2026-07-29T09:59:00Z"
    }
  ],
  "minCoverageRatio": 0,
  "windowEnd": "2026-07-29T10:00:00Z"
}
Expected outputjson
{
  "windowResults": [
    {
      "name": "fast",
      "start": "2026-07-29T09:00:00Z",
      "end": "2026-07-29T10:00:00Z",
      "durationSeconds": 3600,
      "errorRatio": 0.1,
      "burnRate": 10,
      "threshold": 14.4,
      "level": "PAGE",
      "coverageRatio": 0.9,
      "pointCount": 3,
      "verdict": "NOT_TRIPPED"
    },
    {
      "name": "fast_confirm",
      "start": "2026-07-29T09:55:00Z",
      "end": "2026-07-29T10:00:00Z",
      "durationSeconds": 300,
      "errorRatio": 0.1,
      "burnRate": 10,
      "threshold": 14.4,
      "level": "PAGE",
      "coverageRatio": 0.4,
      "pointCount": 2,
      "verdict": "NOT_TRIPPED"
    },
    {
      "name": "medium",
      "start": "2026-07-29T04:00:00Z",
      "end": "2026-07-29T10:00:00Z",
      "durationSeconds": 21600,
      "errorRatio": 0.1,
      "burnRate": 10,
      "threshold": 6,
      "level": "TICKET",
      "coverageRatio": 0.15,
      "pointCount": 3,
      "verdict": "TRIPPED"
    },
    {
      "name": "slow",
      "start": "2026-07-26T10:00:00Z",
      "end": "2026-07-29T10:00:00Z",
      "durationSeconds": 259200,
      "errorRatio": 0.1,
      "burnRate": 10,
      "threshold": 1,
      "level": "TICKET",
      "coverageRatio": 0.0125,
      "pointCount": 3,
      "verdict": "TRIPPED"
    }
  ],
  "alertLevel": "BUDGET_EXHAUSTED",
  "errorBudget": {
    "objective": 0.99,
    "periodDays": 30,
    "budgetRatio": 0.01,
    "budgetEvents": 30,
    "consumedBudgetRatio": 10,
    "remainingBudgetRatio": 0
  },
  "summary": {
    "windowEnd": "2026-07-29T10:00:00Z",
    "seriesPointCount": 3,
    "inWindowPointCount": 3,
    "scale": 6,
    "roundingMode": "HALF_UP"
  },
  "burnRate": 10,
  "dataSufficiency": "SUFFICIENT"
}
Tip
A ten percent error ratio against a one percent budget is a burn rate of ten, which clears neither page threshold of 14.4 but does clear both ticket thresholds, and since that rate has already spent the thirty-day budget the exhausted budget outranks every window verdict.

A short series that refuses to extrapolate

Inputjson
{
  "windows": [
    {
      "name": "fast",
      "durationSeconds": 3600,
      "level": "PAGE"
    }
  ],
  "slo": {
    "periodDays": 30,
    "objective": 0.99
  },
  "series": [
    {
      "errorRatio": 0.5,
      "t": "2026-07-29T09:58:00Z"
    }
  ],
  "windowEnd": "2026-07-29T10:00:00Z"
}
Expected outputjson
{
  "windowResults": [
    {
      "name": "fast",
      "start": "2026-07-29T09:00:00Z",
      "end": "2026-07-29T10:00:00Z",
      "durationSeconds": 3600,
      "errorRatio": 0.5,
      "burnRate": 50,
      "threshold": 14.4,
      "level": "PAGE",
      "coverageRatio": 0,
      "pointCount": 1,
      "verdict": "INSUFFICIENT_DATA"
    }
  ],
  "alertLevel": "BUDGET_EXHAUSTED",
  "errorBudget": {
    "objective": 0.99,
    "periodDays": 30,
    "budgetRatio": 0.01,
    "budgetEvents": null,
    "consumedBudgetRatio": 50,
    "remainingBudgetRatio": 0
  },
  "summary": {
    "windowEnd": "2026-07-29T10:00:00Z",
    "seriesPointCount": 1,
    "inWindowPointCount": 1,
    "scale": 6,
    "roundingMode": "HALF_UP"
  },
  "burnRate": 50,
  "dataSufficiency": "INSUFFICIENT"
}
Tip
One point spans no time at all, so coverage is zero and the window reports insufficient data instead of a fifty-times burn rate.
  • Decide whether a burn justifies a page or only a ticket, with the AND applied
  • Report how much of a 30-day error budget a period has consumed

Avoid for

  • Computing the indicator itself, which observability.calculate_sli answers
  • Summarising raw latency samples, which observability.summarize_metrics_window answers

Composable with

Safety & privacy

  • Runs without calling any external service.
  • Your payloads are never stored permanently.
  • Telemetry captures metadata only — never your payload contents.
  • We never train models on your payloads.
  • No human reviews your payloads.

Tags

analysissreobservabilitymetrics