Reflect — session reflection

Analyze current session and propose additions to the project backlog.

1. Find backlog target

Check memory for saved preference:

node "${CLAUDE_PLUGIN_ROOT}/scripts/memory.js" search "backlog_target"

If no saved preference — detect automatically:

1. Check common locations: backlog.md, TODO.md, TODO, docs/backlog.md
2. Check if project uses GitHub Issues: gh issue list --limit 1 2>/dev/null
3. If nothing found — ask:

Where do you track backlog?
  1. backlog.md (will create)
  2. TODO.md
  3. GitHub Issues (gh issue create)
  4. Other — specify path or tool

Save preference:

node "${CLAUDE_PLUGIN_ROOT}/scripts/memory.js" note "CONFIG: backlog_target=<chosen path or github-issues>"

Read current backlog to avoid duplicates.

1b. Collect watcher auto-findings

The session watcher captures decisions, plans, corrections, and phase changes in real time. Review them before analyzing manually — many candidates for backlog / DOC notes are already surfaced:

node "${CLAUDE_PLUGIN_ROOT}/scripts/memory.js" findings

Output groups entries by type (decision, plan, correction, phase). Treat them as candidates, not authoritative:

• correction — strong candidate for feedback/ memory
• decision — candidate for decisions/ or backlog
• plan — candidate for backlog entry
• phase — context only, usually skip

User confirms which findings promote to backlog / memory in step 3.

2. Analyze session

Review what happened and find:

• Workarounds — bugs/problems worked around, not fixed
• Repetitions — pattern that appeared >1 time, asks for automation
• Gaps — skill/rule that was missing, had to do manually
• Insights — decisions worth scaling to other areas
• Done — what from backlog was completed this session

DOC classification check

For each insight, ask: "Should any contributor to this project know this?" If yes — it's a DOC, not just a backlog item or feedback. Save it immediately:

node "${CLAUDE_PLUGIN_ROOT}/scripts/memory.js" note "DOC: <domain> — <insight>"

Examples of what IS a DOC:

• Tool/framework choices: "use node:test, not vitest" — project convention
• Architecture decisions: "services return null, not throw" — pattern everyone must follow
• Build/deploy constraints: "no external deps in hooks" — technical requirement

Examples of what is NOT a DOC:

• Personal workflow preference: "I prefer short commit messages"
• One-time workaround: "had to restart dev server"
• Backlog item: "refactor auth module"

3. Propose items

Show list in format:

## Reflect: <date>

### New items
- **<section>**: <description> — <why, with specific case from session>

### Close
- <item that is done>

### Skip
- (nothing found / everything already in backlog)

Do NOT write yet. Show user, wait for confirmation.

4. After confirmation

Update the backlog target:

• File-based (backlog.md, TODO.md, etc.): add items to appropriate section, mark closed with [x]
• GitHub Issues: gh issue create --title "<item>" --body "<why>" for each new item; gh issue close <number> for closed

Also save insights to memory (included in the same confirmation — no separate prompt needed):

node "${CLAUDE_PLUGIN_ROOT}/scripts/memory.js" note "REFLECT: <key insight from session>"

5. Check DOC: notes

node "${CLAUDE_PLUGIN_ROOT}/scripts/memory.js" docs

If DOC: notes exist — notify user: "Found N documentation findings. Run /docs-reflect to process them, or /session-end will handle it."

If called from /session-end — pass control to /docs-reflect automatically.

6. Parked items that matured

node "${CLAUDE_PLUGIN_ROOT}/scripts/memory.js" parked

If a parked item became a backlog item, a decision or got done in this session — offer its exit (unpark <file> --to=<where>, after moving the content). Don't push old ones that didn't move.

Rules

• Don't duplicate what's already in backlog
• Quality over quantity — 1 good item beats 5 vague ones
• Wording: what to do + why (specific case from session)
• Done items stay in their section with [x], don't move to separate block