Weekly Engineering Retrospective Pack
Weekly Engineering Retrospective Pack
ClaudeChatGPT
Health
0/100
▲ 23
📋 145 copies
Trigger Phrase
Run the retro pack
Prompt
130 wordsA set of prompts to run your weekly retro in under 30 minutes.
Step 1 — Data Gather (before the meeting)
Prompt: I'll give you our sprint data. Summarise: what shipped, what didn't, and what blocked us. Data: [PASTE JIRA/LINEAR EXPORT OR BULLET POINTS]
Step 2 — What Went Well
Prompt: Based on this sprint summary: [SUMMARY]. Generate 5 specific 'what went well' candidates. Be specific — not 'good communication' but 'the daily standups caught the API issue 2 days early'.
Step 3 — What to Improve
Prompt: Generate 5 specific improvement candidates. For each one, suggest a concrete change we could make next sprint, not a vague principle.
Step 4 — Action Items
Prompt: Given these improvement areas: [LIST]. Write 3 action items. Each must have: a specific owner role, a due date (use 'by next Friday' etc), and a measurable outcome.
Step 1 — Data Gather (before the meeting)
Prompt: I'll give you our sprint data. Summarise: what shipped, what didn't, and what blocked us. Data: [PASTE JIRA/LINEAR EXPORT OR BULLET POINTS]
Step 2 — What Went Well
Prompt: Based on this sprint summary: [SUMMARY]. Generate 5 specific 'what went well' candidates. Be specific — not 'good communication' but 'the daily standups caught the API issue 2 days early'.
Step 3 — What to Improve
Prompt: Generate 5 specific improvement candidates. For each one, suggest a concrete change we could make next sprint, not a vague principle.
Step 4 — Action Items
Prompt: Given these improvement areas: [LIST]. Write 3 action items. Each must have: a specific owner role, a due date (use 'by next Friday' etc), and a measurable outcome.
Install Instructions
Run each step in your preferred AI. Step 1 needs real sprint data — export from your project tool. Steps 2-4 use the output of Step 1.
Test It
Test command:
Run Step 1 with a fake sprint summary
Expected output:
Should produce a structured summary with shipped/blocked/blocker categories
Pass criteria:
- Output has clear categories. Blockers are specific, not generic.
⚠️ Guardrails
- Don't share personal performance data about individuals with AI tools.