Improve

• If found → suggest merging into a single memory file.

After health check, proceed to feedback analysis.

Step 1 — Collect feedback memories

Glob memory/*.md. For each file, read the YAML frontmatter. Keep only files where type: feedback.

If no feedback memories exist, stop:

No feedback memories found in memory/.
Feedback memories are created automatically when you correct Claude during a session.
Nothing to improve yet — keep working and check back later.

Step 2 — Group by area

Read the content of each feedback memory. Extract the area/topic from:

• The description field in frontmatter
• Keywords in the content (testing, imports, lint, naming, formatting, security, etc.)

Group feedback files by area. Count entries per area.

Step 3 — Filter by argument

If $ARGUMENTS is not empty, filter to only the matching area. If no match, list available areas and ask which one.

If $ARGUMENTS is empty, continue with all areas.

Step 4 — Analyze and recommend

For each area, apply thresholds:

Print a summary:

=== Feedback Analysis ===

<area>: <N> feedback entries → <recommendation>
<area>: <N> feedback entries → <recommendation>
<area>: <N> feedback entries → Skip (not a pattern yet)

Which area do you want to act on? (or "all" for batch processing)

If only one area qualifies (3+), print the recommendation and proceed directly to Step 5 without asking.

Step 5 — Pre-flight checks

Before acting, verify constraints:

1. Read .claude/rules/common-mistakes.md — count current items

   • If at 10 items and action is "add to common-mistakes" → suggest /graduate first
   • If at 8-9 items → warn that capacity is almost full

2. Glob .claude/rules/*.md — estimate total token budget

   • If approaching ~500 tokens → warn before creating a new rule file

3. Check if a control already exists for this area

   • Glob .claude/rules/ for files matching the area keyword
   • If found → suggest updating the existing control instead of creating a new one

Step 6 — Execute action

For common-mistakes (3-4 feedback entries)

1. Synthesize the feedback entries into a single, concise mistake description

2. Show the proposed entry to the user:

   Proposed common-mistakes entry:
   
   N. **<Title>** — <What goes wrong and the correct approach>
   
   Source: <N> feedback memories (<list file names>)
   
   Add this to common-mistakes.md?
   

3. Wait for user approval. If rejected, stop — leave memories unchanged.

4. Edit .claude/rules/common-mistakes.md — append the new numbered item

5. Proceed to Step 7 (cleanup)

For controls (5+ feedback entries)

1. Synthesize feedback entries into a control description

2. Show the proposed control to the user:

   Proposed control from <N> feedback entries:
   
   Name: <area>-conventions
   Surface: <rule / permissions / hook / hybrid>
   Scope: <global or path-specific if rule is involved>
   Content:
     - <directive 1>
     - <directive 2>
     - ...
   
   Create this control? (I'll use /new-rule to scaffold it properly)
   

3. Wait for user approval. If rejected, stop — leave memories unchanged.

4. Provide the control details so the user can invoke /new-rule <name> with context, or directly apply changes following /new-rule surface classification

Step 7 — Clean up resolved feedback

After successfully creating a control or common-mistake entry:

1. List the feedback memory files that were consumed

2. Ask the user:

   These feedback memories are now captured as a <control/common-mistake>:
     - memory/<file1>.md
     - memory/<file2>.md
     - ...
   
   Delete them? (They're redundant now — the rule enforces the behavior every session)
   

3. If yes → delete the files and remove their entries from memory/MEMORY.md

4. If no → leave them (user may want to keep for reference)

Step 8 — Print summary

=== Done ===

Action: <Added to common-mistakes.md / Created control <name>>
Source: <N> feedback memories from <area>
Cleaned up: <N files deleted / no files deleted>

Remaining feedback areas with patterns:
  <area>: <N> entries (run /improve <area> to process)