| name | update-learning-logs |
| description | Update the project learning logs (PowerShell/cmd, Git/GitHub/Copilot, AWS, TypeScript, Next.js/React, Playwright) by merging new knowledge into existing categories across one or more files. |
Goal
Keep learning notes searchable by topic (not by date) and continuously improved via merge updates (統合アップデート) instead of duplicating entries.
When to use
Use this skill when:
- We executed or discussed PowerShell/cmd commands.
- We executed or discussed Git/GitHub operations (branches, PRs, Actions, Copilot, MCP).
- We executed or discussed AWS operations (S3/CloudFront/OAC/IAM/AWS CLI).
- We implemented or discussed TypeScript / Next.js / React topics.
- We added or modified Playwright tests, configuration, or debugging steps.
Target files (choose one or more)
| Topic | File |
|---|---|
| PowerShell/cmd | docs/command-learning-log.md |
| Git/GitHub/Copilot/MCP | docs/git-github-learning-log.md |
| AWS | docs/aws-learning-log.md |
| TypeScript | docs/typescript-learning-log.md |
| Next.js/React | docs/nextjs-learning-log.md |
| Playwright | docs/playwright-learning-log.md |
Process (multi-file aware)
- Collect new learnings from the current work (commands used, concepts explained, errors encountered, decisions made).
- Decide target logs (one or multiple files) based on topic.
- For each target file:
- Locate the existing best-matching category section.
- Merge-update the existing entry:
- Prefer adding missing sections (基礎/関連/応用/トラブル) over creating a new duplicate entry.
- If there are duplicates, consolidate into one canonical section and keep the best wording/examples.
- Keep headings stable so future updates land in the same place.
- Add a short
学んだ日: YYYY-MM-DDline when new knowledge was added.
- Show the proposed diffs (file list + what changed) and ask for confirmation if committing.
- If the user requests commit/push:
- Stage only the learning-log related files (avoid accidental staging of generated artifacts like
playwright-report/,test-results/, etc.). - Provide a clear commit message (e.g.,
docs: update learning logs (git/github + playwright)), then push.
- Stage only the learning-log related files (avoid accidental staging of generated artifacts like
Output checklist
- Which log files were updated (may be multiple)
- What knowledge was added (bullets)
- Any risky commands/topics called out
- VERIFY commands to re-check
- Rollback steps (how to revert)