@windagency/valora 2.0.1

package/data/commands/feedback.md

@@ -0,0 +1,600 @@
+---
+name: feedback
+description: Capture outcomes and user feedback to refine agents and prompts through continuous learning and workflow optimization
+experimental: true
+argument-hint: '[--command=<command-name>] [--pr=<pr-number>] [--satisfaction=<1-10>] [--interactive] [--metrics] [--suggest-improvements] [--export=<format>] [--no-stash]'
+allowed-tools:
+  - codebase_search
+  - read_file
+  - grep
+  - list_dir
+  - glob_file_search
+  - run_terminal_cmd
+  - query_session
+  - mcp_github
+model: gpt-5-thinking-high
+agent: product-manager
+stash_protection: true
+prompts:
+  pipeline:
+    - stage: context-identify
+      prompt: context.identify-completed-workflow
+      required: true
+      inputs:
+        command_name: $ARG_command
+        pr_number: $ARG_pr
+      outputs:
+        - workflow_executed
+        - commands_chain
+        - start_time
+        - end_time
+        - execution_duration
+    - stage: context-analyze
+      prompt: context.analyze-command-execution
+      required: true
+      inputs:
+        workflow_executed: $STAGE_context-identify.workflow_executed
+        commands_chain: $STAGE_context-identify.commands_chain
+      outputs:
+        - agents_used
+        - prompts_executed
+        - tools_invoked
+        - files_changed
+        - errors_encountered
+        - retries_performed
+    - stage: context-git
+      prompt: context.gather-git-metrics
+      required: true
+      parallel: true
+      inputs:
+        pr_number: $ARG_pr
+      outputs:
+        - commits_created
+        - files_modified
+        - lines_changed
+        - test_coverage_delta
+        - ci_status
+        - review_comments_count
+    - stage: review-feedback
+      prompt: review.collect-user-feedback
+      required: false
+      conditional: interactive_mode == true || satisfaction_arg != null
+      inputs:
+        workflow_executed: $STAGE_context-identify.workflow_executed
+        satisfaction_arg: $ARG_satisfaction
+        interactive_mode: $ARG_interactive
+      outputs:
+        - satisfaction_score
+        - feedback_comments
+        - improvement_suggestions
+        - pain_points
+        - success_highlights
+    - stage: review-performance
+      prompt: review.calculate-performance-metrics
+      required: true
+      inputs:
+        execution_duration: $STAGE_context-identify.execution_duration
+        commands_chain: $STAGE_context-identify.commands_chain
+        errors_encountered: $STAGE_context-analyze.errors_encountered
+        retries_performed: $STAGE_context-analyze.retries_performed
+      outputs:
+        - time_efficiency_score
+        - error_rate
+        - completion_success_rate
+        - bottlenecks_identified
+        - performance_trends
+    - stage: review-quality
+      prompt: review.evaluate-quality-outcomes
+      required: true
+      inputs:
+        files_changed: $STAGE_context-analyze.files_changed
+        test_coverage_delta: $STAGE_context-git.test_coverage_delta
+        ci_status: $STAGE_context-git.ci_status
+        review_comments_count: $STAGE_context-git.review_comments_count
+      outputs:
+        - code_quality_score
+        - test_quality_score
+        - review_quality_score
+        - overall_quality_score
+    - stage: review-improvements
+      prompt: review.identify-improvement-areas
+      required: true
+      inputs:
+        agents_used: $STAGE_context-analyze.agents_used
+        prompts_executed: $STAGE_context-analyze.prompts_executed
+        bottlenecks_identified: $STAGE_review-performance.bottlenecks_identified
+        errors_encountered: $STAGE_context-analyze.errors_encountered
+      outputs:
+        - agent_improvements
+        - prompt_refinements
+        - workflow_optimizations
+        - tool_suggestions
+        - training_recommendations
+    - stage: summary
+      prompt: documentation.generate-feedback-summary
+      required: true
+      inputs:
+        workflow_executed: $STAGE_context-identify.workflow_executed
+        execution_duration: $STAGE_context-identify.execution_duration
+        satisfaction_score: $STAGE_review-feedback.satisfaction_score
+        time_efficiency_score: $STAGE_review-performance.time_efficiency_score
+        error_rate: $STAGE_review-performance.error_rate
+        completion_success_rate: $STAGE_review-performance.completion_success_rate
+        bottlenecks_identified: $STAGE_review-performance.bottlenecks_identified
+        overall_quality_score: $STAGE_review-quality.overall_quality_score
+        code_quality_score: $STAGE_review-quality.code_quality_score
+        test_quality_score: $STAGE_review-quality.test_quality_score
+        review_quality_score: $STAGE_review-quality.review_quality_score
+        agent_improvements: $STAGE_review-improvements.agent_improvements
+        prompt_refinements: $STAGE_review-improvements.prompt_refinements
+        workflow_optimizations: $STAGE_review-improvements.workflow_optimizations
+      outputs:
+        - feedback_summary
+        - key_insights
+        - recommendations
+  merge_strategy: sequential
+  rollback_on_failure: context
+  cache_strategy: stage
+  retry_policy:
+    max_attempts: 2
+    backoff_ms: 500
+    retry_on:
+      - error
+---
+
+# Feedback Command
+
+## Role
+
+Use the [agent] profile
+
+## Goal
+
+**Capture comprehensive feedback and outcomes** from AI-assisted development workflows to enable continuous improvement of agents, prompts, and processes. This command provides systematic feedback collection, performance analysis, quality assessment, and actionable recommendations for optimizing the development workflow.
+
+### Success Criteria
+
+- ✅ Workflow execution chain is accurately reconstructed
+- ✅ Performance metrics are comprehensively measured (time, errors, retries)
+- ✅ Quality outcomes are objectively assessed (code, tests, reviews)
+- ✅ User satisfaction is captured (if interactive mode)
+- ✅ Bottlenecks and pain points are identified
+- ✅ Specific improvement recommendations are generated
+- ✅ Agent and prompt refinements are suggested
+- ✅ Feedback report is stored in knowledge base
+- ✅ Patterns and best practices are extracted
+- ✅ Metrics are exported for analysis (if requested)
+
+## Rules
+
+### Feedback Collection
+
+1. **MUST reconstruct workflow execution**:
+   - Identify all commands executed
+   - Extract agent/prompt pairs used
+   - Capture timing information
+   - Map command dependencies
+
+2. **MUST gather comprehensive metrics**:
+   - Git and repository metrics (commits, files, lines) — use structured `git` output
+   - CI/CD status and test results
+   - Performance metrics (time, errors, retries)
+   - Quality metrics (code, tests, reviews)
+
+3. **SHOULD collect user feedback** (when possible):
+   - Use `--interactive` for guided feedback session
+   - Accept `--satisfaction=<1-10>` for quick rating
+   - Capture pain points and success highlights
+
+### Performance Assessment
+
+1. **MUST calculate objective scores**:
+   - Time efficiency score (0-100)
+   - Error rate and retry metrics
+   - Completion success rate
+   - Quality scores (code, test, review)
+
+2. **MUST identify bottlenecks**:
+   - Commands taking >150% of baseline time
+   - High-error operations
+   - User-reported slow parts
+
+### Improvement Identification
+
+1. **MUST generate actionable recommendations**:
+   - Agent-specific improvements
+   - Prompt refinements
+   - Workflow optimizations
+   - Tool suggestions
+
+2. **MUST prioritize by impact**:
+   - High: >10 min savings or critical issues
+   - Medium: 5-10 min savings or quality improvements
+   - Low: <5 min savings or nice-to-haves
+
+### Output (No File Generation)
+
+**IMPORTANT: This command does NOT write files to the .valora folder.**
+
+1. **MUST display summary in terminal**:
+   - Executive summary (2-3 sentences)
+   - Performance and quality metrics
+   - Prioritized recommendations
+   - Key insights
+
+2. **Insights are for display only**:
+   - Patterns and learnings shown in output
+   - No files created in .valora folder
+   - No knowledge base modifications
+
+### Git Safety (CRITICAL)
+
+**IMPORTANT: This command MUST NOT modify the git working tree.**
+
+1. **MUST stash uncommitted changes** at start (unless `--no-stash`):
+   - Check for uncommitted changes: `git status --porcelain`
+   - If changes exist, ask user: "Uncommitted changes detected. Stash them before gathering metrics?"
+   - If user agrees, run: `git stash push -m "ai-feedback-auto-stash"`
+   - Record that stash was created
+
+2. **MUST restore changes** at end:
+   - If stash was created, run: `git stash pop`
+   - Verify working tree is restored
+
+3. **NEVER run destructive git commands**:
+   - ❌ `git checkout .` or `git checkout <file>`
+   - ❌ `git reset` (any form)
+   - ❌ `git restore`
+   - ❌ `git clean`
+   - ❌ `git stash drop` (without user consent)
+   - ✅ Only read operations: `git log`, `git diff`, `git status`, `git show`, `git rev-list`
+
+4. **If git commands fail**:
+   - Report the error to the user
+   - Do NOT attempt to "fix" git state
+   - Continue with available data
+
+## Context
+
+### User-Provided Arguments
+
+```plaintext
+$ARGUMENTS
+```
+
+### Available Arguments
+
+#### Command Context
+
+- `--command=<command-name>` - Specify the last command executed (e.g., `create-pr`, `implement`)
+- `--pr=<pr-number>` - Link feedback to specific pull request number
+- `--workflow=<workflow-name>` - Specify full workflow (e.g., `feature-implementation`, `bugfix`)
+
+#### Satisfaction Input
+
+- `--satisfaction=<1-10>` - Quick satisfaction rating (1=very unsatisfied, 10=very satisfied)
+- `--comment="<text>"` - Free-form feedback comment
+- `--interactive` - Enter interactive feedback mode with guided questions
+
+#### Analysis Options
+
+- `--metrics` - Generate detailed performance and quality metrics (enabled by default)
+- `--suggest-improvements` - Enable AI-powered improvement suggestions (enabled by default)
+- `--compare-baseline` - Compare metrics against historical baseline
+- `--identify-bottlenecks` - Highlight slowest operations (enabled by default)
+
+#### Export Options
+
+- `--export=<format>` - Export feedback report (formats: `markdown`, `json`, `csv`, `html`)
+- `--output=<path>` - Custom output path for report
+- `--summary-only` - Generate summary report only (skip detailed analysis)
+
+## Integration with Workflow
+
+### Position in Development Lifecycle
+
+The feedback command typically runs after workflow completion to capture learnings:
+
+```plaintext
+create-pr → feedback → [next-task]
+```
+
+### Typical Usage Patterns
+
+#### Pattern 1: After PR Merge
+
+```bash
+create-pr
+# PR gets reviewed and merged
+feedback --pr=<number> --interactive
+```
+
+#### Pattern 2: After Complete Workflow
+
+```bash
+fetch-task → plan → implement → test → commit → create-pr
+feedback --command=create-pr --satisfaction=8
+```
+
+#### Pattern 3: Quick Feedback
+
+```bash
+feedback --satisfaction=9 --comment="Great workflow, very smooth!"
+```
+
+#### Pattern 4: Detailed Analysis with Export
+
+```bash
+feedback --pr=456 --interactive --export=json --metrics
+```
+
+## Examples
+
+### Example 1: Quick Feedback After PR Creation
+
+**Input:**
+
+```bash
+feedback --command=create-pr --satisfaction=9
+```
+
+**Output:**
+
+```plaintext
+📊 Collecting feedback for workflow: create-pr
+
+✅ Workflow Analysis Complete
+- Duration: 3 minutes
+- Agent: @lead
+- Files changed: 12
+- Quality score: 92/100
+
+📝 Feedback Report Saved
+- Path: .valora/feedback/2025-11-15-1430-create-pr.md
+- Satisfaction: 9/10 (Excellent!)
+
+💡 Key Insights:
+✅ PR creation was fast and high quality
+✅ All checks passed, reviewers assigned successfully
+
+No significant improvements identified. Great workflow! 🎉
+```
+
+### Example 2: Interactive Feedback Session
+
+**Input:**
+
+```bash
+feedback --interactive --pr=456
+```
+
+**Output:**
+
+```plaintext
+📊 Feedback Session: Feature Implementation Workflow
+
+Workflow Overview:
+- Commands: fetch-task → plan → implement → test → commit → create-pr
+- Duration: 1h 30m
+- PR #456: feat(auth): add OAuth2 authentication
+- Status: ✅ Merged
+
+---
+
+Overall, how satisfied are you with this workflow? (1-10): 8
+
+✨ What went well?
+> Implementation was smooth, AI understood requirements clearly.
+
+😤 What was frustrating or slow?
+> Test phase took 15 minutes. Expected 5 minutes.
+
+💡 What could be improved?
+> Better linting before commit, faster tests.
+
+⏱️ Which part took longer than expected?
+> The test command.
+
+❓ Were there any confusing steps?
+> No.
+
+🎯 Any suggestions for specific agents or prompts?
+> Engineer agent should check linting before finishing.
+
+---
+
+📈 Analyzing Performance Metrics...
+
+Performance Summary:
+- Time Efficiency: 75/100
+- Error Rate: 4% (3 errors, 2 auto-resolved)
+- Completion Success: 100% ✅
+
+Quality Summary:
+- Code Quality: 88/100 ✅
+- Test Quality: 82/100 ✅
+- Review Quality: 90/100 ✅
+- Overall: 87/100 ✅
+
+🔍 Identified Bottlenecks:
+1. Test execution (15 min vs 5 min expected)
+   → Parallelize test suites, optimize setup
+
+2. Manual linter fixes (2 retries)
+   → Add real-time linting validation
+
+💡 Improvement Recommendations:
+
+High Priority:
+1. Optimize Test Performance (Save 10 min/workflow)
+2. Add Real-Time Linter Validation (Save 5 min)
+
+Medium Priority:
+3. Parallel Execution (Save 5 min)
+
+---
+
+✅ Feedback Report Generated
+📄 Path: .valora/feedback/2025-11-15-1430-feature-implementation.md
+
+🧠 Knowledge Base Updated:
+- 2 patterns learned
+- 2 best practices added
+- Agent profiles refined
+
+Thank you for your feedback! 🙏
+```
+
+### Example 3: Feedback with JSON Export
+
+**Input:**
+
+```bash
+feedback --pr=789 --export=json
+```
+
+**Output:**
+
+```plaintext
+📊 Analyzing Workflow from PR #789...
+
+Workflow: Bugfix Implementation
+Duration: 45 minutes
+
+✅ Analysis Complete
+
+Performance: 82/100 ✅
+Quality: 90/100 ✅
+
+💡 3 improvements identified
+
+📄 Reports Generated:
+- Markdown: .valora/feedback/2025-11-15-1600-bugfix-implementation.md
+- JSON: .valora/feedback/2025-11-15-1600-bugfix-implementation.json
+
+🧠 Knowledge Base Updated
+```
+
+## Best Practices
+
+### ✅ DO
+
+- **Provide feedback regularly** - After each significant workflow
+- **Be specific** - Mention specific commands, agents, or steps
+- **Rate honestly** - Honest feedback drives real improvements
+- **Use interactive mode** - Provides richer, more actionable insights
+- **Review feedback reports** - Check generated reports for accuracy
+
+### ❌ DON'T
+
+- **Skip feedback** - Feedback drives continuous improvement
+- **Provide vague comments** - "It was fine" doesn't help
+- **Only provide negative feedback** - Highlight successes too
+- **Ignore recommendations** - Review and consider AI suggestions
+
+## Error Handling
+
+### No Recent Workflow Detected
+
+```plaintext
+⚠️  No recent workflow detected.
+
+Please specify a workflow using:
+- --command=<command-name>
+- --pr=<pr-number>
+
+Or run feedback immediately after completing a workflow.
+```
+
+### Git History Unavailable
+
+```plaintext
+❌ Cannot access git history or repository metadata.
+
+Ensure you are in a git repository and have appropriate permissions.
+```
+
+### PR Not Found
+
+```plaintext
+❌ Pull request #123 not found.
+
+Please verify the PR number and repository access.
+```
+
+## Performance Considerations
+
+- **Fast feedback collection** - Basic feedback in < 30 seconds
+- **Detailed analysis** - With `--metrics`, may take 1-2 minutes
+- **Interactive mode** - Depends on user response time
+- **Export operations** - Large JSON exports may take a few seconds
+
+## Document Generation
+
+**This command does NOT generate files.** All output is displayed in the terminal.
+
+## Command Output Summary
+
+Print the following summary at command completion:
+
+**For complete feedback:**
+
+```markdown
+## ✅ Feedback Collected
+
+**Workflow**: [workflow-name]
+**Satisfaction**: [X]/10
+**Duration**: [X] minutes
+
+### Performance Metrics
+
+| Metric          | Score    |
+| --------------- | -------- |
+| Time Efficiency | [XX]/100 |
+| Error Rate      | [X]%     |
+| Completion Rate | [XX]%    |
+
+### Quality Metrics
+
+| Metric         | Score    |
+| -------------- | -------- |
+| Code Quality   | [XX]/100 |
+| Test Quality   | [XX]/100 |
+| Review Quality | [XX]/100 |
+
+### Key Insights
+
+- [Insight 1]
+- [Insight 2]
+
+### Recommendations
+
+1. **[Category]**: [Recommendation] (Impact: [High/Med/Low])
+2. **[Category]**: [Recommendation] (Impact: [High/Med/Low])
+
+### Next Step
+
+→ `/fetch-task` for next task
+→ Project complete! 🎉 (if all tasks done)
+```
+
+**For quick feedback:**
+
+```markdown
+## ✅ Quick Feedback Recorded
+
+**Workflow**: [workflow-name]
+**Satisfaction**: [X]/10
+
+### Summary
+
+[Brief feedback captured]
+
+### Document Generated
+
+→ `.valora/feedback/[timestamp]-[workflow].md`
+
+### Next Step
+
+→ `/fetch-task` for next task
+```