@litfamily/litopencode 1.0.15 → 1.0.17

package/skills/lit-pptx/references/examples/05-research-talk-paper-en.md

@@ -0,0 +1,162 @@
+---
+tonality: paper
+title: Low-temperature CO2 methanation on Ni-Ce catalysts
+subtitle: Activity, stability and the role of ceria oxygen vacancies
+date: 2026-10-06
+department: Example Institute of Catalysis
+presenter: Example research group
+notice: Sample data for illustration only
+---
+layout: cover-index
+
+# Low-temperature CO2 methanation on Ni-Ce catalysts
+---
+layout: section-rule
+
+# Background and method
+---
+layout: text-column
+
+## Research question and scope
+
+- **Question** whether ceria loading lowers the light-off temperature of Ni catalysts below 250 °C
+- **Why it matters** grid-coupled methanation units cycle often, and cold starts waste hydrogen
+- **Scope** five catalysts, 10 wt% Ni on Al2O3 with 0-20 wt% CeO2
+- **Conditions** H2/CO2 = 4, GHSV 24,000 mL g-1 h-1, 0.1 MPa, 150-400 °C
+- **Not covered** sulfur poisoning and pressurised operation above 1 MPa
+- Source: Example group lab notebook, March-August 2026
+---
+layout: method
+
+## Rate expression and fitted terms
+
+r = k · p_CO2^a · p_H2^b / (1 + K_H2O · p_H2O)
+
+- **r** CO2 consumption rate
+  (1) mol per gram of surface Ni per second
+  (2) measured below 10% conversion so the bed stays differential
+- **k** rate constant
+  (1) Arrhenius form, k = A · exp(−Ea / RT)
+  (2) apparent Ea compared across the five catalysts
+- **a, b** reaction orders in CO2 and H2
+  (1) a between 0.21 and 0.34, b between 0.48 and 0.62
+  (2) fitted at fixed temperature, varying one partial pressure at a time
+- **K_H2O** water adsorption term
+  (1) accounts for product inhibition above 220 °C
+  (2) held at one value per catalyst across temperatures
+- **Fit** least squares over 42 runs at 180-260 °C
+  (1) R² = 0.97, residuals without trend in temperature
+- Note: rates are normalised by surface Ni from H2 chemisorption
+- Source: Example group kinetic dataset, runs 101-142
+---
+layout: figure-academic
+
+## Reactor setup and catalyst bed
+
+![Fixed-bed quartz reactor with online gas chromatograph, 6 mm inner diameter](assets/example-a.png)
+
+- 100 mg catalyst diluted with 400 mg SiC to keep the bed isothermal
+- Thermocouple in the bed centre, gradient under 2 K at full conversion
+- Products analysed every 12 minutes by GC with TCD and FID
+- Source: Example group reactor log, 2026
+---
+layout: section-rule
+
+# Results
+---
+layout: chart-insight
+
+## CO2 conversion by temperature and ceria loading
+
+::: chart type=line unit="%"
+| Temperature | 0 wt% Ce | 5 wt% Ce | 10 wt% Ce | 20 wt% Ce |
+|---|---|---|---|---|
+| 175 °C | 2 | 4 | 9 | 7 |
+| 200 °C | 6 | 13 | 27 | 21 |
+| 225 °C | 17 | 34 | 58 | 49 |
+| 250 °C | 38 | 61 | 79 | 72 |
+| 275 °C | 62 | 77 | 84 | 81 |
+> CO2 conversion versus temperature for four ceria loadings (sample data)
+:::
+
+- 10 wt% Ce reaches 50% conversion at 219 °C, 45 K below the Ni-only catalyst
+- 20 wt% Ce performs worse than 10 wt%, consistent with Ni sites covered by ceria
+- Source: Example group activity runs, three repeats each
+- Note: equilibrium conversion at 250 °C under these conditions is 93%
+---
+layout: table-insight
+title: bottom-anchor
+
+## Catalyst properties and kinetic parameters
+
+| Catalyst | Ni dispersion | O vacancy ratio | Ea (kJ/mol) | T50 (°C) |
+|---|---|---|---|---|
+| Ni/Al2O3 | 6.1% | 0.00 | 92 | 264 |
+| Ni-5Ce | 7.4% | 0.18 | 84 | 241 |
+| Ni-10Ce | 8.9% | 0.27 | 76 | 219 |
+| Ni-15Ce | 8.2% | 0.25 | 78 | 224 |
+| Ni-20Ce | 7.0% | 0.21 | 81 | 231 |
+
+> Properties of the five catalysts; T50 is the temperature at 50% conversion (sample data)
+
+- Lower Ea tracks the oxygen vacancy ratio across all five samples
+- Dispersion peaks at 10 wt% Ce and falls as ceria starts to cover Ni
+- Source: H2 chemisorption, XPS Ce3+ fraction and fitted kinetics
+---
+layout: figure-pair
+
+## Spent catalyst after 100 hours on stream
+
+![Ni-only catalyst after 100 h, mean Ni particle 18 nm](assets/example-b.png)
+![Ni-10Ce after 100 h, mean Ni particle 9 nm](assets/example-c.png)
+
+- Ceria slows Ni sintering: particle growth of 29% against 140% for the Ni-only catalyst after 100 h at 250 °C, with conversion loss of 3 points against 14
+- Source: TEM counts of 300 particles per sample, stability runs 201-204
+---
+layout: comparison
+
+## Contributions and limits of this study
+
+:::: columns 1fr 1fr
+::: col
+- **Contributions**
+  (1) Activity: T50 lowered by 45 K with 10 wt% Ce
+  (2) Mechanism: Ea linked to the O vacancy ratio
+  (3) Stability: sintering cut by a factor of five
+  (4) Data: kinetic set of 42 runs released
+:::
+::: col
+- **Limits**
+  (1) Activity: tested at 0.1 MPa only
+  (2) Mechanism: vacancy ratio from ex situ XPS
+  (3) Stability: 100 h, short of plant duty cycles
+  (4) Data: one Ni loading, one support
+:::
+::::
+
+- Source: this work; comparison with industrial targets from reference 6
+---
+layout: closing-summary-list
+
+## Conclusions and next experiments
+
+- Ceria at 10 wt% gives the lowest light-off temperature of the five catalysts
+- Apparent activation energy falls in step with the oxygen vacancy ratio
+- Excess ceria covers Ni and costs both dispersion and activity
+- Ceria slows Ni sintering over 100 hours at 250 °C
+- Next: in situ DRIFTS at 200 °C to identify the formate pathway
+- Next: a 500-hour cycling test at 1 MPa with daily cold starts
+---
+layout: references-appendix
+
+## References
+
+- [1] Example A. et al., Ni catalysts for CO2 methanation, Example Journal of Catalysis 412 (2025) 101-115
+- [2] Example B. et al., Ceria promotion in Ni systems, Example Applied Catalysis 330 (2024) 1220-1234
+- [3] Example C. et al., Oxygen vacancies and CO2 activation, Example Surface Science 79 (2025) 55-68
+- [4] Example D. et al., Rate laws for CO2 hydrogenation, Example Chemical Engineering 18 (2023) 301-319
+- [5] Example E. et al., Sintering of supported Ni, Example Materials Letters 64 (2024) 88-96
+- [6] Example F., Power-to-gas reactor requirements, Example Energy Review 9 (2025) 12-40
+- [7] Example G. et al., XPS of Ce3+ and Ce4+, Example Spectroscopy 51 (2022) 433-447
+- [8] Example H. et al., Dynamic operation of methanation units, Example Process Journal 27 (2026) 7-21
+- [9] Example I. et al., Thermodynamic limits of CO2 methanation, Example Thermo Letters 15 (2023) 210-219

package/skills/lit-pptx/references/examples/06-policy-briefing-gazette-ko.md

@@ -0,0 +1,176 @@
+---
+tonality: gazette
+title: 공공 전기차 충전 인프라 확충 방안
+subtitle: 2027~2029년 구별 배치 기준과 재원 분담안
+date: 2026-10-06
+department: 예시시 기후에너지과
+presenter: 친환경교통팀
+notice: 예시 데이터 · 실제 통계와 예산으로 바꿔 쓰는 샘플
+---
+layout: cover-band
+
+# 공공 전기차 충전 인프라 확충 방안
+---
+layout: agenda
+
+## 보고 순서
+
+- **현황** 등록 대수, 충전기 수, 구별 충전기 1기당 차량 수
+- **문제** 공동주택 밀집 지역의 야간 충전 부족과 고장 방치
+- **확충 방안** 배치 기준, 급속·완속 비율, 운영 방식 비교
+- **재원과 일정** 시비·국비·민간 분담과 3개년 설치 일정
+---
+layout: summary-box-list
+
+## 핵심 요약
+
+::: key-message
+2029년까지 공공 충전기 2,400기를 늘려 충전기 1기당 차량 수를 7.8대에서 4.5대로 낮추는 안
+:::
+
+- **현황** 전기차 4.1만 대, 공공 충전기 5,260기, 1기당 7.8대
+  (1) 북부 3개 구는 1기당 11대 이상, 남부 2개 구는 5대 이하
+- **원인** 공동주택 비율 높은 구에 완속 충전기가 부족
+  (1) 야간 충전 수요의 62%가 공동주택 거주자
+- **방안** 완속 1,900기와 급속 500기를 구별 부족분 순으로 배치
+  (1) 고장 72시간 안 수리를 운영 계약 조건에 포함
+- **재원** 총사업비 312억 원, 시비 38%·국비 34%·민간 28%
+  (1) 민간 몫은 운영사 선정 공모로 확보
+- **운영** 민간 운영사 공모, 요금 상한 kWh당 310원
+  (1) 시 직영보다 운영비 연 21억 원 절감 예상
+- **일정** 27년 2월 운영사 선정, 3월 공영주차장부터 착공
+  (1) 첫해에 북부 3개 구 물량의 60% 배치
+- 출처: 예시시 자동차 등록 통계(2026년 8월), 충전기 운영 현황
+---
+layout: ledger-table
+
+## 구별 충전 인프라 현황
+
+| 구 | 전기차 | 공공 충전기 | 1기당 차량 | 공동주택 비율 | 고장률 |
+|---|---|---|---|---|---|
+| 북1구 | 7,840대 | 610기 | 12.9대 | 78% | 6.1% |
+| 북2구 | 6,920대 | 590기 | 11.7대 | 74% | 5.4% |
+| 북3구 | 5,310대 | 470기 | 11.3대 | 69% | 4.8% |
+| 중앙구 | 8,150대 | 1,320기 | 6.2대 | 52% | 3.2% |
+| 남1구 | 7,060대 | 1,410기 | 5.0대 | 47% | 2.7% |
+| 남2구 | 5,820대 | 860기 | 6.8대 | 55% | 3.5% |
+| 합계 | 41,100대 | 5,260기 | 7.8대 | 62% | 3.9% |
+
+> 표 1. 구별 충전 인프라 현황, 2026년 8월 말 기준 (예시 데이터)
+
+- 북부 3개 구는 차량 대비 충전기가 남1구의 절반 수준이고 고장률도 가장 높음
+- 출처: 예시시 자동차 등록 통계, 충전기 운영 현황(2026년 8월)
+- 주: 고장률은 월평균 가동 불가 충전기 비율
+---
+layout: section-band
+
+# 확충 방안
+---
+layout: comparison
+title: side-rail
+
+## 운영 방식 비교
+
+:::: columns 1fr 1fr
+::: col
+- **시 직영**
+  (1) 설치비: 시비와 국비로 전액 부담
+  (2) 운영비: 연 21억 원, 시 예산으로 계속 지출
+  (3) 수리: 위탁 업체 계약, 평균 9일 소요
+  (4) 요금: kWh당 292원으로 시가 결정
+:::
+::: col
+- **민간 운영사 공모**
+  (1) 설치비: 민간이 28% 부담, 나머지 시비·국비
+  (2) 운영비: 운영사 부담, 시는 부지만 제공
+  (3) 수리: 72시간 안 수리를 계약 조건으로 명시
+  (4) 요금: 상한 kWh당 310원 안에서 운영사 결정
+:::
+::::
+
+- 출처: 예시시 충전기 운영비 결산(2025년), 인근 두 도시 공모 사례
+---
+layout: matrix-2x2
+
+## 설치 부지 유형별 우선순위
+
+| 효과 \ 난이도 | 쉬움 | 어려움 |
+|---|---|---|
+| 큼 | 공영주차장 · 완속 620기 · 27년 상반기 | 공동주택 단지 · 완속 900기 · 입주민 동의 필요 |
+| 작음 | 공공청사 · 완속 180기 · 27년 하반기 | 도로변 · 급속 200기 · 전력 인입 공사 필요 |
+
+> 표 2. 설치 부지 유형별 우선순위, 효과는 야간 충전 수요 해소량 기준 (예시)
+
+- 공영주차장과 공동주택 단지가 북부 3개 구 부족분의 71%를 차지
+- 공동주택 단지는 입주민 동의 절차로 착공까지 평균 5개월 소요
+- 도로변 급속 200기는 전력 인입 공사가 끝나는 2028년부터 설치
+- 출처: 예시시 부지 조사(2026년 7~9월)
+---
+layout: table-insight
+title: top-rule
+
+## 연도별 설치 물량과 사업비
+
+| 연도 | 완속 | 급속 | 사업비 | 1기당 차량(목표) |
+|---|---|---|---|---|
+| 2027년 | 700기 | 150기 | 108억 원 | 6.6대 |
+| 2028년 | 650기 | 180기 | 109억 원 | 5.4대 |
+| 2029년 | 550기 | 170기 | 95억 원 | 4.5대 |
+| 합계 | 1,900기 | 500기 | 312억 원 | 4.5대 |
+| 재원 분담 | 시비 119억 원 | 국비 106억 원 | 민간 87억 원 | |
+
+> 표 3. 연도별 설치 물량과 사업비 (예시 데이터)
+
+- 첫해에 북부 3개 구 물량의 60%를 몰아 배치해 1기당 차량 격차부터 줄임
+- 2029년 목표 4.5대는 등록 대수 연 18% 증가를 가정한 값으로, 증가율이 25%면 5.3대
+- 급속 비율은 21%로 유지하고 도로변 급속은 전력 인입 공사가 끝나는 2028년부터 확대
+- 민간 몫 87억 원은 운영사 공모에서 설치비 분담률을 평가 항목으로 넣어 확보
+- 출처: 예시시 3개년 충전 인프라 계획(초안)
+---
+layout: timeline
+
+## 추진 일정
+
+| 시점 | 할 일 | 담당 |
+|---|---|---|
+| 26년 11월 | 시의회 보고와 예산안 반영 | 기후에너지과 |
+| 26년 12월 | 운영사 공모 공고, 평가 기준 공개 | 친환경교통팀 |
+| 27년 2월 | 운영사 선정과 협약 체결 | 계약심사팀 |
+| 27년 3월 | 공영주차장 1차 설치 착공 | 운영사 |
+| 27년 9월 | 1차 설치분 이용률 점검과 2차 배치 조정 | 친환경교통팀 |
+| 27년 12월 | 2027년 850기 설치 완료와 고장 대응 실적 공개 | 운영사 |
+
+> 표 4. 2027년 추진 일정 (예시)
+
+- 예산이 11월에 반영되지 않으면 공모와 착공이 각각 석 달씩 늦어짐
+- 공동주택 단지 설치는 조례 개정 뒤인 27년 하반기부터 가능
+- 이용률이 30%에 못 미치는 지점은 2차 배치에서 물량을 줄여 조정
+- 출처: 예시시 3개년 충전 인프라 계획(초안)
+---
+layout: closing-decision-box
+
+## 건의 사항
+
+- 결정할 것: 민간 운영사 공모 방식과 2027년 사업비 108억 원을 11월 예산안에 반영
+
+| 안건 | 금액 | 결정 시점 | 담당 |
+|---|---|---|---|
+| 2027년 설치 사업비 | 108억 원 | 11월 예산안 | 기후에너지과 |
+| 운영사 공모 방식 채택 | 비용 없음 | 11월 시의회 보고 | 친환경교통팀 |
+| 공동주택 설치 지원 조례 개정 | 비용 없음 | 12월 입법예고 | 법무담당관 |
+
+> 표 5. 결정 요청 안건 (예시)
+---
+layout: references-appendix
+
+## 붙임: 근거 자료와 용어
+
+- [1] 예시시 자동차 등록 통계, 2026년 8월 말 기준 (예시)
+- [2] 예시시 공공 충전기 운영 현황과 고장 기록, 2026년 1~8월 (예시)
+- [3] 예시시 충전기 운영비 결산, 2025년 (예시)
+- [4] 인근 두 도시 충전기 운영사 공모 사례, 2025년 (예시)
+- [5] 예시시 충전기 설치 부지 조사, 2026년 7~9월 (예시)
+- [6] 예시시 3개년 충전 인프라 계획 초안, 2026년 9월 (예시)
+- [7] 1기당 차량: 등록 전기차 수를 공공 충전기 수로 나눈 값
+- [8] 고장률: 월평균 가동 불가 충전기 수를 전체 충전기 수로 나눈 값
+- [9] 완속·급속: 출력 7kW 안팎은 완속, 50kW 이상은 급속으로 구분

package/skills/lit-pptx/references/examples/07-annual-story-studio-en.md

@@ -0,0 +1,131 @@
+---
+tonality: studio
+title: Neighbourhood reading rooms 2026
+subtitle: A year of the network in rooms, readers and hours
+date: 2026-10-06
+department: Example Reading Rooms Trust
+presenter: Programme office
+notice: Sample data and stand-in photos for illustration only
+---
+layout: cover-split-image
+
+# Neighbourhood reading rooms 2026
+---
+layout: asymmetric-feature
+
+## The network at a glance
+
+![Evening study table at the riverside room](assets/example-b.png)
+
+- **Rooms** 23 open, up from 17 in 2025, each within 800 m of a primary school
+- **Hours** 61 hours a week on average, with four rooms open until 22:00
+- **Volunteers** 412 active, giving 38,600 hours across the year
+- **Cost** 1.9M EUR running cost, 82 EUR per registered reader
+- Source: Example Reading Rooms Trust annual accounts, Sep 2025-Aug 2026
+---
+layout: kpi-row
+
+## Readers and visits, September 2025 to August 2026
+
+| Reading rooms | Registered readers | Visits | Loans | Evening visits | Under-16 share |
+|---|---|---|---|---|---|
+| 23 | 23,100 | 318,000 | 146,500 | 34% | 41% |
+| six opened in 2026 | up 28% on 2025 | 13.8 per reader | 6.3 per reader | after 18:00 | of visits |
+
+- Growth came mostly from the six rooms opened in 2026
+- Evening visits rose fastest where rooms stay open past 20:00
+- Source: room check-in system and loan records, full year
+---
+layout: chart-insight
+
+## Visits per month, old rooms and new rooms
+
+::: chart type=column unit="thousand visits"
+| Month | 17 rooms open before 2026 | 6 rooms opened in 2026 |
+|---|---|---|
+| Sep | 21 | 0 |
+| Nov | 22 | 2 |
+| Jan | 23 | 5 |
+| Mar | 22 | 7 |
+| May | 21 | 8 |
+| Jul | 19 | 9 |
+> Visits per month for the established and the new rooms (sample data)
+:::
+
+- New rooms reached 9,000 visits a month within seven months
+- Established rooms held steady; July dips with school holidays
+- Source: room check-in system, monthly totals
+---
+layout: image-full
+
+## Saturday story hour at the market-hall room
+
+![Saturday story hour at the market-hall room, 40 children and carers](assets/example-c.png)
+---
+layout: comparison
+title: side-rail
+
+## Evening hours trial, four rooms
+
+:::: columns 1fr 1fr
+::: col
+- **Before the trial**
+  (1) Hours: closing at 19:00 on weekdays
+  (2) Visits: 11,200 a month across the four rooms
+  (3) Users: 22% aged 16-24
+  (4) Cost: 6,100 EUR a month for staff
+:::
+::: col
+- **During the trial**
+  (1) Hours: closing at 22:00 on weekdays
+  (2) Visits: 15,900 a month across the four rooms
+  (3) Users: 37% aged 16-24
+  (4) Cost: 8,400 EUR a month for staff
+:::
+::::
+
+- Source: trial report, March-August 2026; staff cost from payroll
+---
+layout: photo-grid
+title: bottom-anchor
+
+## Three rooms opened in 2026
+
+![Former bank branch, north district, 140 m²](assets/example-a.png)
+![School annexe, harbour district, 95 m²](assets/example-b.png)
+![Ground floor of a housing block, west district, 120 m²](assets/example-c.png)
+
+- All three were donated or leased for one euro a year for ten years
+- Source: Example Reading Rooms Trust property register
+---
+layout: timeline
+title: top-plain-large
+
+## Plan for the year ahead
+
+| When | What | Who |
+|---|---|---|
+| Nov 2026 | Evening hours extended to eight more rooms | Operations |
+| Jan 2027 | Two rooms opened in the south district | Property team |
+| Mar 2027 | Homework help in all 25 rooms | Programmes |
+| Jun 2027 | Reader survey and room-by-room review | Research |
+| Aug 2027 | Annual accounts and the 2027 review | Finance |
+
+- Evening hours cost about 2,300 EUR per room per month
+- The two south-district rooms open before homework help starts in March
+- Source: Example Reading Rooms Trust plan for 2026-2027
+---
+layout: closing-contact-split
+
+## Support for 2027
+
+- Contact: programme office, rooms@example.org
+
+| Need | Amount | By |
+|---|---|---|
+| Evening hours in eight rooms | 220k EUR | Nov 2026 |
+| Two new rooms in the south | 160k EUR | Jan 2027 |
+| Homework help programme | 90k EUR | Mar 2027 |
+| Book stock renewal | 60k EUR | Jun 2027 |
+
+> Funding sought for 2027 (sample data)

package/skills/lit-pptx/references/examples/08-reliability-keynote-night-en.md

@@ -0,0 +1,182 @@
+---
+tonality: night
+title: Checkout platform reliability, 2026 review
+subtitle: Latency, incidents and error budget across three regions
+date: 2026-10-06
+department: Example Commerce platform engineering
+presenter: Reliability team
+notice: Sample data for illustration only
+---
+layout: cover-figures
+
+# Checkout platform reliability, 2026 review
+---
+layout: kpi-row
+
+## Reliability at a glance, January to September
+
+| Availability | p99 latency | Incidents | Error budget used | Mean time to restore |
+|---|---|---|---|---|
+| 99.962% | 412 ms | 17 | 76% | 38 min |
+| target 99.95% | target 450 ms | 2025: 26 | of 9-month budget | 2025: 61 min |
+
+- Availability met target in eight of nine months; May missed it by 0.04 points
+- Source: Example Commerce SLO dashboard, January-September 2026
+- Note: availability counts successful checkout requests over all checkout requests
+---
+layout: full-chart
+
+## p99 checkout latency by month and region
+
+::: chart type=line unit="ms"
+| Month | EU | US | APAC |
+|---|---|---|---|
+| Jan | 398 | 421 | 512 |
+| Feb | 402 | 418 | 498 |
+| Mar | 391 | 409 | 476 |
+| Apr | 388 | 402 | 455 |
+| May | 446 | 431 | 470 |
+| Jun | 384 | 397 | 441 |
+| Jul | 379 | 392 | 428 |
+| Aug | 381 | 389 | 419 |
+| Sep | 377 | 386 | 412 |
+> p99 latency of the checkout API by region (sample data)
+:::
+
+- APAC fell from 512 ms to 412 ms after the edge cache moved to two local sites in March
+- Source: Example Commerce tracing data, 1% request sample
+---
+layout: dashboard-grid
+
+## Incident profile by cause and severity
+
+::: chart type=bar unit="incidents"
+| Cause | Count |
+|---|---|
+| Config change | 6 |
+| Dependency | 5 |
+| Capacity | 3 |
+| Code defect | 3 |
+> Incidents by root cause (sample data)
+:::
+
+::: chart type=column unit="minutes"
+| Severity | Median restore |
+|---|---|
+| SEV1 | 52 |
+| SEV2 | 37 |
+| SEV3 | 24 |
+> Median minutes to restore by severity (sample data)
+:::
+
+::: chart type=column unit="incidents"
+| Region | Count |
+|---|---|
+| EU | 7 |
+| US | 6 |
+| APAC | 4 |
+> Incidents by region (sample data)
+:::
+
+::: chart type=line unit="incidents"
+| Quarter | Count |
+|---|---|
+| Q1 | 8 |
+| Q2 | 6 |
+| Q3 | 3 |
+> Incidents per quarter (sample data)
+:::
+
+- Config changes caused six of 17 incidents; all six shipped outside the canary path
+- Source: Example Commerce incident register, January-September 2026
+---
+layout: big-number
+title: top-plain-large
+
+## Error budget burn in May
+
+| Budget burned | Duration | Requests failed |
+|---|---|---|
+| 31% | 3 h 40 min | 1.9M |
+| in one incident | SEV1, 14 May | 0.8% of the day |
+
+- One payment-provider timeout cascaded into retries that saturated the EU gateway
+- Retries without backoff tripled load within six minutes
+- Fix: retry budget per client and a circuit breaker on the provider call
+- Source: post-incident review PIR-2026-014
+---
+layout: method
+title: band
+
+## Error budget calculation
+
+budget = (1 − SLO) × total requests in the window
+
+- **SLO** service level objective
+  (1) 99.95% of checkout requests succeed over a rolling 28 days
+  (2) a request fails on a 5xx response or a response slower than 3 s
+- **Window** rolling 28 days
+  (1) about 410M checkout requests at current volume
+  (2) recomputed every five minutes from the request log
+- **Budget** failed requests the window may hold
+  (1) 0.05% of 410M, about 205,000 requests
+  (2) the May incident alone used 31% of one window
+- **Burn rate** speed of spending the budget
+  (1) failed requests per hour divided by the hourly budget
+  (2) a burn rate of 1 spends the budget exactly at the window's end
+- **Alert** paging rule
+  (1) page when the 1-hour burn rate exceeds 14 or the 6-hour rate exceeds 6
+- Source: Example Commerce SLO policy v3, February 2026
+---
+layout: comparison
+
+## Change process before and after canary gating
+
+:::: columns 1fr 1fr
+::: col
+- **Before, January-April**
+  (1) Rollout: config pushed to all regions at once
+  (2) Detection: median 11 minutes after the push
+  (3) Incidents: four from config changes
+  (4) Rollback: manual, median 19 minutes
+:::
+::: col
+- **After, May-September**
+  (1) Rollout: 1% canary for 20 minutes per region
+  (2) Detection: median 4 minutes, inside the canary
+  (3) Incidents: two from config changes
+  (4) Rollback: automatic on burn-rate alert, 3 minutes
+:::
+::::
+
+- Source: Example Commerce change log and incident register, 2026
+---
+layout: timeline
+
+## Reliability roadmap, Q4 2026 to Q2 2027
+
+| When | What | Owner |
+|---|---|---|
+| Oct 2026 | Canary gating for feature flags | Platform |
+| Nov 2026 | Retry budgets in all client SDKs | Payments |
+| Jan 2027 | Second APAC region active-active | Infra |
+| Mar 2027 | Chaos test of provider failover | Reliability |
+| May 2027 | SLO raised to 99.97% | Product |
+
+- Raising the SLO halves the error budget, so it waits for the APAC region
+- The March chaos test checks failover before the SLO goes up in May
+- Source: Example Commerce platform roadmap, October 2026
+---
+layout: closing-decision-box
+
+## Decisions for the Q4 roadmap
+
+- Decide: fund the second APAC region and adopt retry budgets as a release gate
+
+| Decision | Cost | When | Owner |
+|---|---|---|---|
+| Second APAC region | 0.9M USD a year | 20 Oct | Infra |
+| Retry budget as a gate | none | 27 Oct | Payments |
+| Flag canary gating | 2 engineer-months | 3 Nov | Platform |
+
+> Decisions requested (sample data)