---
id: review.identify-improvement-areas
version: 1.0.0
category: review
experimental: true
name: Identify Improvement Areas
description: Generate actionable recommendations for agents, prompts, and workflow optimization
tags:
  - improvement
  - recommendations
  - optimization
  - continuous-learning
model_requirements:
  min_context: 128000
  recommended:
    - claude-sonnet-4.6
    - claude-sonnet-5
agents:
dependencies:
  requires:
    - context.use-modern-cli-tools
    - context.analyze-command-execution
    - review.calculate-performance-metrics
    - review.collect-user-feedback
inputs:
  - name: agents_used
    description: List of agents used in workflow
    type: array
    required: true
  - name: prompts_executed
    description: List of prompts executed
    type: array
    required: true
  - name: bottlenecks_identified
    description: Performance bottlenecks
    type: array
    required: true
  - name: pain_points
    description: User-reported pain points
    type: array
    required: false
  - name: errors_encountered
    description: Errors during execution
    type: array
    required: true
outputs:
  - agent_improvements
  - prompt_refinements
  - workflow_optimizations
  - tool_suggestions
  - training_recommendations
tokens:
  avg: 3500
  max: 7000
  min: 1800
---

# Identify Improvement Areas

## Objective

Analyze workflow execution data, performance metrics, and user feedback to generate specific, actionable recommendations for improving agents, prompts, and processes.

## Instructions

### Step 1: Analyze Agent Performance

For each agent in `agents_used`:

**Performance indicators:**

- Execution time (from bottlenecks)
- Error count (from errors_encountered)
- User satisfaction (from pain_points)
- Retry rate

**Classification:**

- **High-performing**: Fast, low errors, positive feedback
- **Adequate**: Meets expectations
- **Needs improvement**: Slow, frequent errors, or negative feedback

**Generate recommendations:**

- Configuration adjustments
- Specialization refinements
- Tool additions
- Context improvements

### Step 2: Analyze Prompt Effectiveness

For each prompt in `prompts_executed`:

**Effectiveness indicators:**

- Errors during prompt execution
- Retries required
- Output quality (inferred from subsequent steps)
- User feedback on prompt outputs

**Issue patterns:**

- High error rate → Prompt lacks validation
- Frequent retries → Prompt unclear or missing context
- Poor quality → Prompt instructions insufficient

**Generate refinements:**

- Add clarifications
- Include additional context
- Improve validation rules
- Refine output format

### Step 3: Identify Workflow Optimizations

**Analyze workflow for:**

1. **Parallelization opportunities:**
   - Which commands can run concurrently?
   - Example: Run test and review-code in parallel

2. **Conditional skipping:**
   - Which steps can be skipped based on context?
   - Example: Skip assert if coverage >90%

3. **Caching opportunities:**
   - What data can be cached between commands?
   - Example: Cache codebase analysis from plan to implement

4. **Decision point adjustments:**
   - Are thresholds appropriate?
   - Do loops terminate efficiently?

5. **New command patterns:**
   - Repeated sequences that could be automated
   - Common workflows needing dedicated commands

**Estimate time savings:**

- Calculate potential reduction in duration
- Quantify efficiency gains

### Step 4: Suggest Tool Improvements

Based on bottlenecks and pain points:

**Missing tools:**

- What operations are manual that could be automated?
- Example: Auto-fix linter errors

**Tool enhancements:**

- Existing tools that need improvements
- Example: Faster test execution

**Integration opportunities:**

- External tools to integrate
- Example: Code complexity analyzer

### Step 5: Generate Training Recommendations

**Knowledge gaps identified:**

- Agents lacking context
- Prompts with repeated patterns
- Common error scenarios

**Training needs:**

- Update agent profiles
- Add examples to prompts
- Document best practices
- Create new knowledge base entries

## Output Format

```json
{
	"agent_improvements": [
		{
			"agent": "backend-engineer-api",
			"performance": "good",
			"issues_found": [],
			"suggestions": [
				"Add real-time linting validation to prevent post-implementation fixes",
				"Include error handling patterns by default"
			],
			"priority": "medium",
			"estimated_impact": "Save ~5 minutes per workflow"
		},
		{
			"agent": "qa",
			"performance": "needs_improvement",
			"issues_found": ["Test execution took 15 minutes (expected: 5 minutes)"],
			"suggestions": ["Optimize test setup/teardown", "Parallelize test suites", "Cache test dependencies"],
			"priority": "high",
			"estimated_impact": "Save ~10 minutes per workflow"
		}
	],
	"prompt_refinements": [
		{
			"prompt": "code.implement-changes",
			"issues_found": ["Generated code had 2 linter errors"],
			"suggestion": "Add instruction: 'Validate code against linter rules before outputting'",
			"priority": "high",
			"estimated_impact": "Reduce errors by 50%"
		},
		{
			"prompt": "code.implement-tests",
			"issues_found": [],
			"suggestion": "Add instruction: 'Optimize test performance, use mocks for external dependencies'",
			"priority": "medium",
			"estimated_impact": "Improve test speed"
		}
	],
	"workflow_optimizations": [
		{
			"type": "parallelization",
			"suggestion": "Run 'test' and 'review-code' in parallel after implementation",
			"rationale": "These commands are independent",
			"estimated_time_saved": "5 minutes",
			"priority": "high",
			"implementation_complexity": "low"
		},
		{
			"type": "conditional_skip",
			"suggestion": "Skip 'assert' step if test coverage is >90%",
			"rationale": "High coverage indicates completeness",
			"estimated_time_saved": "2 minutes",
			"priority": "medium",
			"implementation_complexity": "low"
		},
		{
			"type": "caching",
			"suggestion": "Cache codebase analysis results from 'plan' phase for use in 'implement'",
			"rationale": "Avoid redundant file scanning",
			"estimated_time_saved": "3 minutes",
			"priority": "medium",
			"implementation_complexity": "medium"
		}
	],
	"tool_suggestions": [
		{
			"tool": "auto_fix_linter",
			"type": "new",
			"rationale": "Reduce manual linter error fixes",
			"impact": "Save ~5 minutes per workflow",
			"priority": "high"
		}
	],
	"training_recommendations": [
		"Update @qa agent with test performance optimization patterns",
		"Add linter validation examples to @engineer agent training",
		"Document common error patterns in knowledge base",
		"Create best practice guide for test optimization"
	]
}
```

## Success Criteria

- ✅ All agents analyzed with performance classification
- ✅ Prompt effectiveness assessed
- ✅ Workflow optimizations identified
- ✅ Improvements prioritized by impact
- ✅ Estimated time savings calculated
- ✅ Recommendations are specific and actionable

## Error Handling

- **Insufficient data**: Mark recommendations as "low confidence"
- **No issues found**: Return positive feedback, skip suggestions
- **Conflicting recommendations**: Prioritize by user pain points