praveenmaiya/weekly-update

Weekly Update Skill

Generate a concise, business-focused weekly status update for the team.

Usage: /weekly-update Jan 20

Process

Step 1: Gather Context

1. Read STATUS_LOG.md - find entries within the 7-day window ending on the given date
2. Run git log for the period:

   git log --oneline --since="7 days ago" --until="tomorrow"
   

   (Adjust dates based on the provided date argument)
3. Check doc timestamps:

   ls -la docs/*.md
   

4. Read docs/weekly_updates.md - match the existing format and tone

Step 2: Draft the Update

Write 2-3 bullets following these rules:

| Rule | Do This | |------|---------| | Work, not stats | Describe what was accomplished; stats support the story, don't lead it | | Business impact | Why does this matter? What does it unlock? | | Self-explanatory | Reader should understand without a voiceover | | No "we" or "I" | Noun-led phrasing ("Established..." not "I established...") | | No version numbers | Say "segment-based recommendations" not "v5.17" | | Human tone | Conversational, clear, not robotic or AI-sounding | | Brief | 2-3 bullets maximum |

Step 3: Ask for "Next"

IMPORTANT: Always ask the user:

"What's coming next week for the Next section?"

Wait for user response before proceeding. Do not auto-generate the Next section.

Step 4: Present Draft

Show the complete update:

**Weekly Update - [Date], 2026**

* **[Topic]:** What was done. Why it matters.
* **[Topic]:** ...
* **Next:** [user-provided next items]

Step 5: Offer to Save

Ask: "Want me to add this to docs/weekly_updates.md and commit?"

If yes:

1. Prepend the update to docs/weekly_updates.md with --- separator
2. Commit with message: Add weekly update [Date]: [brief summary]

Anti-patterns

| Don't Write | Write Instead | |-------------|---------------| | "Implemented collaborative filtering" | "Collaborative filtering was tested but deprioritized due to low repeat purchase rates (18%)" | | "Fixed variant dedup bug" | "Rolled out a fix for specialized product variants, ensuring users see exact performance parts" | | "87% coverage achieved" | "Vehicle-specific recommendations now reach 87% of users" | | "We investigated the issue" | "Investigation revealed that 65% of purchases involve untracked products" | | "v5.17 deployed" | "Segment-based recommendations deployed" |

Example Output

Weekly Update - Jan 20, 2026

• Segment-Based Recommendations Validated (Holley): The shift from global popularity to vehicle-segment recommendations is now live and measured. Open rates nearly tripled and click-through rates nearly doubled—same-user testing confirmed it's the algorithm, not seasonality.
• Deployment Verification Runbooks: Established standardized guides to trace a model from "Succeeded" workflow to live serving config. Cuts debugging time on silent failures where workflows pass but models don't update.
• Next: Run end-to-end training flow for a test company. Start scoping generalized recommendation framework.