+ Clearly separates the push failure from successful payment.
- Omits the retry-payment attempts and double-charge inquiries.
a senior software analyst. ## Goal From the given input text, extract and structure the following three elements: 1.
| Category | Development › Coding |
|---|---|
| Tags | SummarizingReformattingDeveloperCode |
Act as a senior software analyst.
## Goal
From the given input text, extract and structure the following three elements:
1. describ_feature → What feature or system is being discussed
2. what_should_happen → Expected behavior
3. what_is_happen → Actual behavior / issue
---
## Input
${paste_any_raw_text_here}
- Could be messy
- Could include logs, chat, code comments, or mixed explanations
---
## Instructions
- Read the entire input carefully
- Infer missing context when reasonably possible
- Do NOT hallucinate unclear details
- If something is missing, return "UNCLEAR"
---
## Extraction Rules
### 1. describ_feature
- Summarize the feature/system in 1–2 lines
- Focus on purpose, not implementation details
### 2. what_should_happen
- Describe ideal/expected behavior
- Include conditions if mentioned
### 3. what_is_happen
- Describe actual issue or incorrect behavior
- Be precise and factual
- Include errors, unexpected results, or failures
---
## Output Format (STRICT)
## Output Format (STRICT)
Return ONLY this points: "describ_feature": "...",
"what_should_happen": "...",
"what_is_happen": "..."
---
## Constraints
- No extra text
- No explanations
- No assumptions beyond reasonable inference
- Keep each field concise but completeUsed to turn messy bug-related text into a simple structured report before filing or debugging. Unclear items are marked UNCLEAR, and it outputs only the three specified fields.
ChatGPT is the most concise and precise, while Gemini adds fuller user impact. [C] is missing and therefore ranks last.
+ Clearly separates the push failure from successful payment.
- Omits the retry-payment attempts and double-charge inquiries.
+ Covers the cause, retry failure, and user impact.
- The actual-behavior field is slightly long and impact-heavy.
| Criterion | ChatGPT | Gemini | Leader |
|---|---|---|---|
| Instruction following | 10 | 10 | Tie |
| Accuracy | 10 | 10 | Tie |
| Specificity | 10 | 10 | Tie |
| Structure | 10 | 10 | Tie |
| Right length | 10 | 9 | ChatGPT +11% |
Scored 1–10 by gpt-5.6-sol with model names hidden (2026-09-25). This is an AI review, not a measurement.
We gave three models the same input and copied their answers unedited. Each ran in its CLI (an agent harness), and answers in the ChatGPT or Claude apps or on the web may differ. Outputs are in Korean.
[Pasted material] 제목: [버그] 결제 완료 푸시 알림 미발송 건 작성자: 김도영 (백엔드팀, ShopEasy) 안녕하세요, 어제(9/23)부터 결제 완료 푸시 알림이 오지 않는다는 CS가 5건 접수됐습니다. 정상 흐름: 주문이 PG사 결제 승인 콜백을 받으면 OrderCompleteListener가 동작해서 3초 이내에 사용자 앱으로 '결제가 완료되었습니다' FCM 푸시를 보내야 합니다. 로그: 2026-09-23 14:02:11 [ERROR] PushService: FCM send failed, reason=SenderId mismatch 2026-09-23 14:02:11 [WARN] OrderCompleteListener: retry exceeded (3/3), giving up 2026-09-23 14:02:12 [INFO] OrderService: order#88213 status=PAID (push not sent) 어제 오전에 FCM 서버 키를 새로 발급받아 로테이션했는데, 그 이후로 전송이 전부 실패하고 있습니다. 결제 자체는 정상 처리되고 주문 상태도 PAID로 잘 바뀌는데, 알림만 안 가는 상황이라 사용자들이 결제가 됐는지 몰라서 재결제를 시도하다가 이중결제 문의까지 들어왔습니다.
"describ_feature": "PG 결제 승인 후 사용자 앱에 결제 완료 FCM 푸시 알림을 발송하는 시스템",
"what_should_happen": "결제 승인 콜백 수신 후 OrderCompleteListener가 3초 이내에 '결제가 완료되었습니다' 푸시 알림을 보내야 합니다.",
"what_is_happen": "FCM 서버 키 로테이션 이후 SenderId mismatch 오류로 푸시 전송이 전부 실패하고, 3회 재시도 후 중단되고 있습니다. 결제와 주문 상태 변경은 정상이나 알림이 발송되지 않습니다."
| 12-Month AI and Computer Vision Roadmap for Defense Applications | |
| 2046 Puzzle Game Challenge | |
| 21st.dev component prompt | |
| 3D FACTORY | |
| 3D FPS Game |