npm - claude-flow-novice - Versions diffs - 2.18.6 → 2.18.7 - Mend

claude-flow-novice 2.18.6 → 2.18.7

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (1000) hide show

package/claude-assets/skills/cfn-ruvector-codebase-index/SKILL.md DELETED Viewed

@@ -1,830 +0,0 @@
----
-name: cfn-ruvector-codebase-index
-description: Semantic codebase indexing and search using RuVector vector database
-version: 1.0.0
-author: Claude Flow Novice
-category: search
-tags: [ruvector, search, semantic, indexing, embeddings]
----
-# RuVector Codebase Index Skill
-Semantic codebase indexing and search using RuVector's vector database. Enables natural language queries to find relevant files based on meaning, not just keywords.
-## Features
-### Codebase Intelligence
-- **Full Reindexing**: Rebuild entire codebase index from scratch
-- **Incremental Updates**: Update index for specific files only
-- **Auto-Indexing**: Automatic updates on git commits (via hook)
-- **Semantic Search**: Natural language queries using OpenAI embeddings
-- **Stale Documentation Detection**: Identify legacy/outdated .md files automatically
-- **Multi-Language Support**: TypeScript, JavaScript, Python, Rust, Go, Java, C/C++, C#, Ruby, PHP, Swift, Kotlin, Markdown
-- **Metadata Extraction**: Parses exports, dependencies, purpose, complexity
-### Error Pattern Learning (New)
-- **Error Storage**: Store failed task patterns with context and solutions
-- **Error Query**: Search for relevant past failures before starting similar work
-- **Automatic Avoidance**: Learn from mistakes to prevent repeating errors
-### Learning & Best Practices (New)
-- **Pattern Storage**: Store successful patterns and best practices (PATTERN, STRAT, ANTI, EDGE)
-- **Learning Query**: Search for relevant learnings before implementation
-- **Knowledge Accumulation**: Build organizational knowledge base over time
-- **Confidence Tracking**: Tag learnings with confidence scores and metadata
-### Cost Optimization
-- **Z.ai Support**: Use Z.ai provider for low-cost embeddings ($0.50/1M tokens)
-## Quick Start
-### 1. Initial Setup
-```bash
-# Install dependencies (if not already installed)
-cd docker/trigger-dev
-npm install
-# Set API key (choose one)
-export OPENAI_API_KEY=your-key    # OpenAI
-export ZAI_API_KEY=your-key       # Z.ai (cost-optimized)
-```
-### 2. Index Changed Files (Recommended Default)
-```bash
-# Via slash command (auto-detects git changes)
-/codebase-reindex
-# Or directly
-./.claude/skills/cfn-ruvector-codebase-index/index.sh --auto
-```
-**For full rebuild (only when needed):**
-```bash
-# Explicitly request full reindex
-./.claude/skills/cfn-ruvector-codebase-index/index.sh --full
-```
-**Expected output:**
-```
-[INFO] Starting full reindex...
-[INFO] Found 1247 files to index
-[PROGRESS] Indexing 1247/1247 files...
-[SUCCESS] Full reindex completed
-[INFO] Indexed: 1242 files
-[WARN] Failed: 5 files
-```
-### 3. Search
-```bash
-# Via slash command
-/codebase-search authentication logic --top 10
-# Or directly
-./.claude/skills/ruvector-codebase-index/search.sh "authentication logic" --top 10
-```
-**Expected output:**
-```
-═══════════════════════════════════════════════════════
-   Search Results for: "authentication logic"
-═══════════════════════════════════════════════════════
-[0.85]  src/auth/AuthContext.tsx
-    Purpose: React context for authentication state management
-    Exports: AuthProvider, useAuth, AuthContext
-    Lines: 145, Complexity: 12
-[0.78]  src/lib/auth-utils.ts
-    Purpose: Authentication utility functions and helpers
-    Exports: validateToken, refreshAuth, logout
-    Lines: 89, Complexity: 8
-```
-### 4. Enable Auto-Indexing (Optional)
-```bash
-# Install git hook
-./.claude/skills/ruvector-codebase-index/install-hook.sh
-# Now every commit automatically updates the index!
-```
-## Architecture
-```
-┌─────────────────────────────────────────────────────────┐
-│                  User Interactions                       │
-├─────────────────────────────────────────────────────────┤
-│  /codebase-reindex        → Full rebuild from scratch   │
-│  /codebase-search <query> → Semantic search             │
-│  git commit               → Auto-incremental update     │
-└─────────────────────────────────────────────────────────┘
-                           ↓
-┌─────────────────────────────────────────────────────────┐
-│              RuVector Codebase Index Skill               │
-├─────────────────────────────────────────────────────────┤
-│  index.sh      → Index files (full or incremental)      │
-│  search.sh     → Query semantic index                   │
-│  embeddings.js → Generate OpenAI embeddings             │
-│  parser.js     → Extract file metadata (AST)            │
-└─────────────────────────────────────────────────────────┘
-                           ↓
-┌─────────────────────────────────────────────────────────┐
-│            RuVector Database (codebase_index)            │
-├─────────────────────────────────────────────────────────┤
-│  Per File Entry:                                         │
-│  - text: Combined content for embedding                  │
-│  - metadata: {filePath, exports, dependencies, ...}      │
-│  - vector: 1536-dim embedding (OpenAI ada-002)          │
-└─────────────────────────────────────────────────────────┘
-```
-## File Structure
-```
-.claude/skills/ruvector-codebase-index/
-├── SKILL.md              # This file
-├── config.json           # Configuration
-├── index.sh              # Main indexing script
-├── search.sh             # Search interface
-├── embeddings.js         # OpenAI embedding generator
-├── parser.js             # AST-based metadata extractor
-└── install-hook.sh       # Git hook installer
-.claude/commands/
-├── codebase-reindex.md   # Slash command: /codebase-reindex
-└── codebase-search.md    # Slash command: /codebase-search <query>
-.claude/hooks/
-└── post-commit-codebase-index  # Git hook for auto-indexing
-```
-## Configuration
-**File:** `config.json`
-```json
-{
-  "indexableExtensions": [".ts", ".tsx", ".js", ".jsx", ".py", ...],
-  "ignorePatterns": ["node_modules/**", "dist/**", ...],
-  "maxFileSize": 1048576,
-  "embeddingModel": "text-embedding-3-small",
-  "embeddingDimensions": 1536,
-  "batchSize": 100,
-  "ruvectorDbPath": "./docker/trigger-dev/data"
-}
-```
-**Customization:**
-- Add/remove file extensions in `indexableExtensions`
-- Add ignore patterns to skip specific directories
-- Adjust `maxFileSize` to index larger files
-- Change `batchSize` for different rate limiting needs
-## Usage Patterns
-### Quick Decision Guide: Which Index Mode?
-```
-┌─────────────────────────────────────────────────────────────┐
-│ Choose Your Indexing Mode                                   │
-├─────────────────────────────────────────────────────────────┤
-│ ❓ Regular workflow / After code changes?                   │
-│ ❓ Have uncommitted changes to index?                       │
-│ ❓ Want to index all git-tracked changes?                   │
-│ ❓ Just committed and need to update index?                 │
-│    └─→ USE: --auto (DEFAULT, RECOMMENDED)                   │
-│       Command: /codebase-reindex (or index.sh --auto)       │
-│       Time: 5-30 seconds (only changed files)               │
-│                                                              │
-│ ❓ First time setup?                                        │
-│ ❓ Major codebase restructure?                              │
-│ ❓ Config changed (extensions, ignore patterns)?            │
-│ ❓ Index corrupted or missing?                              │
-│    └─→ USE: --full (EXPLICIT ONLY)                          │
-│       Command: index.sh --full                              │
-│       Time: 2-30 minutes (all 8k+ files)                    │
-│                                                              │
-│ ❓ Modified specific files manually?                        │
-│ ❓ Want to index new files without full rebuild?            │
-│ ❓ Debugging index issues for certain files?                │
-│    └─→ USE: --files <path1> <path2> ...                    │
-│       Command: index.sh --files src/auth.ts src/utils.ts   │
-│       Time: 1-2 seconds per file                            │
-│                                                              │
-│ ❓ Want automatic updates on every commit?                  │
-│ ❓ Never want to think about indexing?                      │
-│    └─→ USE: Git hook (automatic, background)               │
-│       Setup: install-hook.sh (one-time)                     │
-│       Behavior: Auto-indexes on git commit                  │
-└─────────────────────────────────────────────────────────────┘
-```
-**Quick Reference:**
-| Mode | Command | When | Speed |
-|------|---------|------|-------|
-| **Auto (Default)** | `/codebase-reindex` | Regular workflow, after commits | 5-30s |
-| **Full** | `index.sh --full` | First setup, corruption, config change | 2-30 min |
-| **Specific** | `index.sh --files <paths>` | Manual file updates | 1-2s/file |
-| **Hook** | `install-hook.sh` (once) | Every commit (automatic) | Background |
----
-### Full Reindex (From Scratch)
-**When to use:**
-- First-time setup
-- Major codebase restructuring
-- Index appears corrupted
-- After config changes
-```bash
-./.claude/skills/ruvector-codebase-index/index.sh --full
-```
-**What it does:**
-1. Deletes existing `codebase_index.db`
-2. Scans entire project for indexable files
-3. Parses metadata for each file
-4. Generates embeddings (OpenAI/Z.ai)
-5. Stores in RuVector database
-**Performance:**
-- ~1000 files: 2-3 minutes
-- ~5000 files: 10-15 minutes
-- Rate limited to respect API quotas
-### Incremental Update (Specific Files)
-**When to use:**
-- Updated specific files manually
-- Want to index new files without full rebuild
-- Debugging index issues for specific files
-```bash
-./.claude/skills/ruvector-codebase-index/index.sh --files src/auth/AuthContext.tsx src/lib/utils.ts
-```
-**What it does:**
-1. Parses only specified files
-2. Generates embeddings
-3. Updates/inserts into existing database
-**Performance:** ~1-2 seconds per file
-### Auto-Detection (Git Changes)
-**When to use:**
-- Want to index all uncommitted changes
-- Manually trigger incremental update
-- Testing before commit
-```bash
-./.claude/skills/ruvector-codebase-index/index.sh --auto
-```
-**What it does:**
-1. Detects staged and modified files via `git diff`
-2. Filters for indexable extensions
-3. Incrementally updates index
-### Semantic Search
-**Examples:**
-```bash
-# Find authentication code
-/codebase-search authentication logic
-# Find React components
-/codebase-search React components for user profile --top 10
-# Find database utilities
-/codebase-search database migration utilities
-# Find error handling
-/codebase-search error handling patterns
-```
-**Search Tips:**
-- Use natural language descriptions
-- Be specific about functionality (not file names)
-- Increase `--top` for more comprehensive results
-- Results are ranked by semantic similarity
-### Stale Documentation Detection (NEW!)
-**When to use:**
-- Regular documentation hygiene (monthly/quarterly)
-- Before major refactors
-- When cleaning up legacy code
-- Identifying docs to archive
-```bash
-# Via slash command
-/detect-stale-docs
-# Or directly
-./.claude/skills/ruvector-codebase-index/detect-stale-docs.sh
-```
-**What it detects:**
-1. **Missing References**: Files/functions mentioned in docs that no longer exist
-   ```
-   [STALE] Score: 15 | Age: 180d
-     File: docs/OLD_AUTH_SYSTEM.md
-     - Missing file: src/auth/OldAuthProvider.tsx
-     - Missing file: src/lib/legacy-auth.ts
-     - HIGH: 75% of file references are missing
-   ```
-2. **Orphaned Documentation**: Docs with no code references
-   ```
-   [LIKELY STALE] Score: 8 | Age: 120d
-     File: planning/prototype-ideas.md
-     - No code references found (orphaned)
-   ```
-3. **Deprecated Keywords**: Docs mentioning "legacy", "deprecated", "obsolete"
-   ```
-   [STALE] Score: 12 | Age: 200d
-     File: docs/LEGACY_DEPLOYMENT.md
-     - Contains deprecated/legacy keywords
-   ```
-4. **Age Factor**: Old docs more likely to be stale
-   ```
-   [POSSIBLY STALE] Score: 4 | Age: 365d
-     File: docs/api-reference-v1.md
-   ```
-**Scoring System:**
-| Score | Status | Action |
-|-------|--------|--------|
-| >= 10 | **STALE** | Strong candidate for archival |
-| 5-9 | **LIKELY STALE** | Needs review and updates |
-| 2-4 | **POSSIBLY STALE** | Minor issues, low priority |
-| < 2 | Active | No action needed |
-**Score Calculation:**
-- Age: +1 per 90 days
-- Missing file refs: +1 per 10% missing
-- Orphaned (no refs): +5
-- Deprecated keywords: +3
-**Expected Output:**
-```
-═══════════════════════════════════════════════════════════════
-   Stale Documentation Report
-═══════════════════════════════════════════════════════════════
-[STALE] Score: 15 | Age: 180d
-  File: ./docs/OLD_AUTH_SYSTEM.md
-  References: 8 files (6 missing), 12 code refs (8 missing)
-  - Missing file: src/auth/OldAuthProvider.tsx
-  - Missing file: src/lib/legacy-auth.ts
-  - HIGH: 75% of file references are missing
-[LIKELY STALE] Score: 8 | Age: 120d
-  File: ./planning/prototype-ideas.md
-  References: 0 files (0 missing), 0 code refs (0 missing)
-  - No code references found (orphaned)
-  - Contains deprecated/legacy keywords
-═══════════════════════════════════════════════════════════════
-Stale docs: 5
-Likely stale: 12
-Total analyzed: 247
-═══════════════════════════════════════════════════════════════
-Recommendations:
-  1. Review files with score >= 10 for archival
-  2. Update files with missing references
-  3. Mark deprecated docs with clear warnings
-  4. Consider moving legacy docs to /archive or /legacy directory
-```
-**Workflow for Cleanup:**
-1. **Run detection:**
-   ```bash
-   /detect-stale-docs
-   ```
-2. **Review high-score files** (>= 10):
-   - Read the file to confirm it's obsolete
-   - Check git history for last meaningful update
-   - Verify code references are truly missing
-3. **Take action:**
-   ```bash
-   # Option A: Archive
-   mkdir -p archive/docs/legacy
-   mv docs/OLD_AUTH_SYSTEM.md archive/docs/legacy/
-   # Option B: Update
-   # Edit the doc to fix references
-   # Option C: Delete
-   git rm docs/completely-obsolete.md
-   ```
-4. **Re-run to verify:**
-   ```bash
-   /detect-stale-docs  # Should show fewer stale docs
-   ```
-**Performance:**
-- ~1-2 seconds per .md file
-- 100 docs: ~2-3 minutes
-- 500 docs: ~10-15 minutes
-- Uses semantic search to verify code references
-## Git Hook Auto-Indexing
-### Installation
-```bash
-./.claude/skills/ruvector-codebase-index/install-hook.sh
-```
-Creates symlink: `.git/hooks/post-commit` → `.claude/hooks/post-commit-codebase-index`
-### Behavior
-**On every commit:**
-1. Extracts list of committed files from `git diff-tree`
-2. Filters for indexable extensions
-3. Triggers incremental index update in **background**
-4. Logs to `/tmp/ruvector-index.log`
-5. **Non-blocking** - commit succeeds even if indexing fails
-### Configuration
-```bash
-# Enable/disable auto-indexing
-export RUVECTOR_AUTO_INDEX=true   # default
-export RUVECTOR_AUTO_INDEX=false  # disable
-# Check logs
-tail -f /tmp/ruvector-index.log
-```
-### Uninstall
-```bash
-rm .git/hooks/post-commit
-```
-## API Integration
-### Using from Scripts
-**Index specific files:**
-```bash
-#!/bin/bash
-files=("src/auth/AuthContext.tsx" "src/lib/utils.ts")
-./.claude/skills/ruvector-codebase-index/index.sh --files "${files[@]}"
-```
-**Search programmatically:**
-```bash
-#!/bin/bash
-RESULTS=$(./.claude/skills/ruvector-codebase-index/search.sh "authentication logic" --top 5)
-echo "$RESULTS" | jq -r '.[].metadata.metadata.filePath'
-```
-### Using from Node.js
-**Direct RuVector access:**
-```javascript
-import { getCollection, COLLECTIONS } from './docker/trigger-dev/src/lib/ruvector-init.ts';
-import { generateEmbedding } from './.claude/skills/ruvector-codebase-index/embeddings.js';
-// Search
-const query = "authentication logic";
-const queryEmbedding = await generateEmbedding(query);
-const collection = getCollection(COLLECTIONS.CODEBASE_INDEX);
-const results = await collection.search({
-  vector: new Float32Array(queryEmbedding),
-  k: 10,
-});
-console.log(results);
-```
-## Performance & Costs
-### Indexing Performance
-| Codebase Size | Full Reindex Time | Incremental Update |
-|---------------|-------------------|--------------------|
-| 100 files     | 20-30 seconds     | 1-2 seconds/file   |
-| 1000 files    | 2-3 minutes       | 1-2 seconds/file   |
-| 5000 files    | 10-15 minutes     | 1-2 seconds/file   |
-| 10000 files   | 20-30 minutes     | 1-2 seconds/file   |
-**Bottlenecks:**
-- Embedding generation (API latency)
-- File I/O (reading file contents)
-- Parsing (TypeScript AST for .ts files)
-**Optimizations:**
-- Batch embedding requests (100 at a time)
-- Parallel file parsing (future enhancement)
-- Caching parsed metadata (future enhancement)
-### Embedding Costs
-**OpenAI `text-embedding-3-small`:**
-- Price: $0.02 / 1M tokens
-- Avg file size: ~500 tokens
-- 1000 files: ~$0.01
-- 10000 files: ~$0.10
-**Z.ai (via API routing):**
-- Price: $0.50 / 1M tokens (25x cheaper than OpenAI)
-- 1000 files: ~$0.0025
-- 10000 files: ~$0.025
-**Recommendation:** Use Z.ai for indexing, OpenAI for critical searches
-### Search Performance
-**Query latency:**
-- Embedding generation: ~200-500ms
-- Vector search: ~10-50ms
-- Total: ~250-600ms
-**Accuracy:**
-- Top-5 recall: ~85-90% (finds relevant files in top 5 results)
-- Top-10 recall: ~95-98%
-## Integration with CFN Loop
-### Using in Agents
-```bash
-# Agent prompt can reference search results
-SEARCH_RESULTS=$(./.claude/skills/ruvector-codebase-index/search.sh "authentication logic" --top 3)
-FILES=$(echo "$SEARCH_RESULTS" | jq -r '.[].metadata.metadata.filePath')
-# Pass to agent
-npx claude-flow-novice agent typescript-specialist \
-  --prompt "Refactor authentication logic in these files: $FILES"
-```
-### Using in Decomposers
-```javascript
-// In cfn-architecture-decomposer.ts
-import { execSync } from 'child_process';
-const searchQuery = "authentication components";
-const searchResults = execSync(
-  `./.claude/skills/ruvector-codebase-index/search.sh "${searchQuery}" --top 5`
-).toString();
-const relevantFiles = JSON.parse(searchResults).map(r => r.metadata.metadata.filePath);
-// Use in micro-task decomposition
-microTasks.push({
-  id: "auth-refactor-1",
-  files: relevantFiles,
-  description: "Refactor authentication logic",
-});
-```
-## Troubleshooting
-### Issue: "RuVector not initialized"
-**Cause:** Database doesn't exist yet
-**Solution:**
-```bash
-# Run full reindex to initialize
-./.claude/skills/ruvector-codebase-index/index.sh --full
-```
-### Issue: "No API key found"
-**Cause:** Missing `OPENAI_API_KEY` or `ZAI_API_KEY`
-**Solution:**
-```bash
-# Add to .env or export
-export ZAI_API_KEY=your-key
-```
-### Issue: "Failed to parse file"
-**Cause:** TypeScript AST parsing error (syntax error in file)
-**Solution:**
-- Check file for syntax errors
-- File will be skipped automatically
-- Review `/tmp/ruvector-index.log` for details
-### Issue: "Search returns no results"
-**Possible causes:**
-1. Index not built yet → Run `/codebase-reindex`
-2. Query too specific → Try broader terms
-3. Files not in indexable extensions → Check `config.json`
-### Issue: "Indexing is slow"
-**Solutions:**
-- Use Z.ai instead of OpenAI (25x cheaper, similar speed)
-- Reduce `batchSize` in config (trades speed for rate limits)
-- Run full reindex overnight for large codebases
-### Issue: "Git hook not triggering"
-**Checks:**
-```bash
-# Verify symlink exists
-ls -la .git/hooks/post-commit
-# Check if RUVECTOR_AUTO_INDEX is disabled
-echo $RUVECTOR_AUTO_INDEX
-# Check logs
-tail -f /tmp/ruvector-index.log
-```
-## Future Enhancements
-### Planned Features
-1. **Parallel Indexing**
-   - Index multiple files concurrently
-   - Target: 5x speedup for full reindex
-2. **Metadata Caching**
-   - Cache parsed metadata to avoid re-parsing
-   - Only regenerate embeddings if file changed
-3. **Incremental Embedding Updates**
-   - Detect if file content changed significantly
-   - Skip re-embedding if only minor changes
-4. **Language-Specific Parsers**
-   - Better metadata extraction for Python, Rust, Go
-   - Currently TypeScript has best support
-5. **Search Filters**
-   - Filter by file type, directory, date
-   - Example: `/codebase-search auth --type ts --dir src/`
-6. **Query Expansion**
-   - Auto-expand queries with synonyms
-   - Example: "auth" → "authentication", "authorization"
-7. **Result Clustering**
-   - Group related files in search results
-   - Example: Show all files in same module together
-## References
-- RuVector Documentation: `docker/trigger-dev/src/lib/ruvector-init.ts`
-- MDAP Integration: `planning/ruvector/MDAP_CEREBRAS_IMPLEMENTER_HANDOFF.md`
-- Schema Definitions: `docker/trigger-dev/src/lib/ruvector-schemas.ts`
-- OpenAI Embeddings: https://platform.openai.com/docs/guides/embeddings
-## Version History
-- **1.0.0** (2025-11-30): Initial implementation
-  - Full reindex, incremental update, auto-indexing
-  - Semantic search with OpenAI/Z.ai embeddings
-  - TypeScript, Python, multi-language support
-  - Git hook integration for auto-updates
----
-**Status:** ✅ Ready for Use | Tested with TypeScript codebases
-**Next Steps:**
-1. Run initial reindex: `/codebase-reindex`
-2. Install git hook: `./.claude/skills/ruvector-codebase-index/install-hook.sh`
-3. Try searching: `/codebase-search your query here`
-## Error Pattern Learning & Best Practices
-### Store Error Pattern
-After a task failure, store the error pattern to avoid repeating it:
-```bash
-./.claude/skills/ruvector-codebase-index/store-error-pattern.sh \
-  --task-id "task-auth-123" \
-  --error-type "TypeScript compilation" \
-  --pattern "Missing type imports in multi-file refactor" \
-  --context "Files: auth.ts, types.ts, middleware.ts. Forgot to add import { User } from './types'" \
-  --solution "Always add type imports before interface usage. Use import type { } syntax."
-```
-### Query Error Patterns
-Before starting similar work, check for past failures:
-```bash
-./.claude/skills/ruvector-codebase-index/query-error-patterns.sh \
-  --task-description "Implement authentication middleware with TypeScript" \
-  --limit 5
-```
-**Output:**
-```
-📚 Relevant error patterns to avoid:
-  ❌ TypeScript compilation: Missing type imports in multi-file refactor
-     Context: Files: auth.ts, types.ts, middleware.ts. Forgot to add import { User } from './types'
-     Solution: Always add type imports before interface usage. Use import type { } syntax.
-     (Task: task-auth-123, 2025-12-01T10:30:00Z)
-  ❌ Runtime error: Middleware ordering issue
-     Context: Auth middleware must run before rate limiter
-     Solution: Check middleware stack order in app setup
-     (Task: task-api-456, 2025-11-28T14:15:00Z)
-```
-### Store Learning/Pattern
-After successful implementation, store the pattern for future reference:
-```bash
-./.claude/skills/ruvector-codebase-index/store-learning.sh \
-  --task-id "task-auth-123" \
-  --category "PATTERN" \
-  --title "Authentication middleware composition pattern" \
-  --description "Middleware stack: validateToken → enrichUserContext → checkPermissions. Each returns next() on success, throws on failure." \
-  --confidence 0.92 \
-  --tags "auth,middleware,express,composition"
-```
-**Categories:**
-- `PATTERN`: Successful implementation patterns to replicate
-- `STRAT`: Strategic approaches that worked well
-- `ANTI`: Anti-patterns to avoid (similar to errors but design-level)
-- `EDGE`: Edge cases and corner case handling
-### Query Learnings
-Before implementation, search for relevant best practices:
-```bash
-./.claude/skills/ruvector-codebase-index/query-learnings.sh \
-  --task-description "Implement authentication middleware with Express" \
-  --category "PATTERN" \
-  --limit 5
-```
-**Output:**
-```
-📚 Relevant learnings/patterns:
-  ✅ [PATTERN] Authentication middleware composition pattern (confidence: 0.92)
-     Middleware stack: validateToken → enrichUserContext → checkPermissions. Each returns next() on success, throws on failure.
-     Tags: auth,middleware,express,composition
-     (Task: task-auth-123, 2025-12-01T10:30:00Z)
-  ✅ [PATTERN] JWT validation best practice (confidence: 0.88)
-     Always verify signature, check expiry, and validate issuer. Use try-catch for decode errors.
-     Tags: jwt,auth,validation,security
-     (Task: task-jwt-789, 2025-11-30T09:45:00Z)
-```
-### Workflow Integration
-**Task Mode - Before Implementation:**
-```bash
-# 1. Query past errors to avoid
-query-error-patterns.sh --task-description "$TASK" --limit 5
-# 2. Query learnings for best practices
-query-learnings.sh --task-description "$TASK" --category "PATTERN" --limit 5
-# 3. Implement with learned knowledge
-# ... (agent execution)
-# 4a. If failed: Store error pattern
-store-error-pattern.sh --task-id "$TASK_ID" --error-type "..." --pattern "..." --solution "..."
-# 4b. If successful: Store learning
-store-learning.sh --task-id "$TASK_ID" --category "PATTERN" --title "..." --description "..." --confidence 0.85
-```
-**Trigger.dev Mode:**
-- Coordinator automatically queries error patterns before decomposition
-- MDAP implementer automatically stores failures
-- Coordinator automatically stores successful patterns after validation