optimization-suggest

1# Optimization Suggest
2 
3Generate data-driven optimization recommendations for Claude Code usage.
4 
5## Input
6 
7The user provides: **$ARGUMENTS**
8 
9This may be:
10- "all" or empty (default: comprehensive optimization scan)
11- "cost" for cost reduction focus
12- "speed" for performance/speed focus
13- "quality" for error reduction focus
14- "efficiency" for workflow efficiency focus
15 
16## Procedure
17 
181. **Gather optimization data** from `http://localhost:4820`:
19   - `GET /api/sessions?limit=200` — session history
20   - `GET /api/analytics` — tool and token analytics
21   - `GET /api/pricing/cost` — cost data
22   - `GET /api/pricing` — pricing rules for model comparison
23   - Sample event streams for behavioral analysis
24 
252. **Analyze optimization opportunities**:
26 
27   ### 💰 Cost Optimization
28   - **Model downgrade opportunities**: Tasks completed with expensive models that could use cheaper ones
29     - Compare success rates per model per task type
30     - Calculate savings from model substitution
31   - **Cache optimization**: Sessions with low cache hit rates
32     - Identify sessions that could benefit from better prompt caching
33   - **Early termination**: Sessions that ran longer than needed
34     - Detect sessions where useful work completed well before session end
35   - **Compaction reduction**: Sessions hitting context limits
36     - Suggest breaking large tasks into smaller sessions
37 
38   ### ⚡ Speed Optimization
39   - **Tool selection**: Faster alternatives for commonly-used tool patterns
40…