UNPKG

claude-flow-novice

Version:

Claude Flow Novice - Advanced orchestration platform for multi-agent AI workflows with CFN Loop architecture Includes Local RuVector Accelerator and all CFN skills for complete functionality.

257 lines (194 loc) 6.24 kB
# CFN Loop Validation Skill ## Overview Advanced validation framework for iterative development workflows with dynamic consensus calculation, multi-mode validation, and adaptive quality assurance. ## Status **OPERATIONAL** - Agent-accessible via CLI wrapper ## Dependencies ### Required - **Node.js** >= 18.0.0 (Runtime for consensus calculator) - **Bash** >= 4.0 (Shell wrapper scripts) - **jq** (JSON parsing and processing) - **SQLite3** (Evidence chain persistence) - **bc** (Basic calculator for threshold comparisons) ### Optional (for enhanced features) - **Redis** (For distributed validation coordination) - Package: `redis-server` and `redis-cli` - Purpose: Publishing validation events to swarm ## Installation ### Install Required Dependencies ```bash # Check Node.js version node --version # Should be >= 18.0.0 # Install jq (JSON processor) # Ubuntu/Debian sudo apt-get install jq bc sqlite3 # macOS brew install jq bc sqlite3 # Windows (WSL) sudo apt-get install jq bc sqlite3 ``` ### Install Optional Dependencies ```bash # Install Redis for distributed coordination # Ubuntu/Debian sudo apt-get install redis-server redis-tools # macOS brew install redis # Start Redis redis-server & ``` ### Verify Installation ```bash # Check Node.js node --version # Should be >= 18.0.0 # Check jq jq --version # Should show version info # Check bc bc --version # Should show version info # Check SQLite sqlite3 --version # Should show version info # Check Redis (optional) redis-cli ping # Should return "PONG" # Test CFN validation CLI ./.claude/skills/cfn-loop-validation/validate-iteration.sh --help ``` ## Quick Start ### Validate Loop 3 Confidence Gate ```bash # MVP mode (low threshold, fast iteration) ./.claude/skills/cfn-loop-validation/validate-iteration.sh \ --mode mvp \ --iteration 1 \ --confidence 0.75 \ --task-id feature-auth # Standard mode (balanced quality) ./.claude/skills/cfn-loop-validation/validate-iteration.sh \ --mode standard \ --iteration 2 \ --confidence 0.85 \ --task-id feature-auth # Enterprise mode (high quality, strict validation) ./.claude/skills/cfn-loop-validation/validate-iteration.sh \ --mode enterprise \ --iteration 1 \ --confidence 0.92 \ --task-id compliance-system ``` ### Validate Loop 2 Consensus ```bash # Validate consensus from validator swarm ./.claude/skills/cfn-loop-validation/validate-iteration.sh \ --mode standard \ --iteration 2 \ --confidence 0.85 \ --consensus 0.92 \ --task-id feature-auth \ --json ``` ### JSON Output for Agent Parsing ```bash # Get JSON output for programmatic use RESULT=$(./.claude/skills/cfn-loop-validation/validate-iteration.sh \ --mode standard \ --iteration 1 \ --confidence 0.85 \ --json) # Parse result echo "$RESULT" | jq -r '.passed' # true/false echo "$RESULT" | jq -r '.status' # PASS/FAIL/MAX_ITERATIONS_EXCEEDED ``` ## Validation Modes | Mode | Gate Threshold | Consensus | Max Iterations | Validators | |------|----------------|-----------|----------------|------------| | **MVP** | ≥0.70 | ≥0.85 | 5 | 2 | | **Standard** | ≥0.80 | ≥0.90 | 10 | 4 | | **Enterprise** | ≥0.90 | ≥0.95 | 15 | 5-8 | ### Mode Selection Guide **MVP Mode:** - Proof of concepts - Rapid prototyping - Learning and experimentation **Standard Mode:** - Production applications - Team collaboration - Quality-focused development **Enterprise Mode:** - Financial systems - Healthcare applications - Compliance-critical systems ## Exit Codes | Code | Status | Meaning | Action | |------|--------|---------|--------| | 0 | PASS | Validation passed | Continue workflow | | 1 | FAIL | Score below threshold | Inject feedback, retry | | 2 | MAX_ITERATIONS_EXCEEDED | Too many attempts | Escalate to human | | 3 | Invalid arguments | Bad CLI input | Fix arguments | | 4 | Configuration error | Config file issue | Check config.json | ## Common Issues ### Issue: "jq: command not found" **Solution:** Install jq ```bash # Ubuntu/Debian sudo apt-get install jq # macOS brew install jq ``` ### Issue: "bc: command not found" **Solution:** Install bc (basic calculator) ```bash # Ubuntu/Debian sudo apt-get install bc # macOS brew install bc ``` ### Issue: "Config file not found" **Solution:** Verify config.json exists ```bash ls -la ./.claude/skills/cfn-loop-validation/config.json ``` ### Issue: "Permission denied" **Solution:** Make script executable ```bash chmod +x ./.claude/skills/cfn-loop-validation/validate-iteration.sh ``` ### Issue: SQLite database errors **Solution:** Ensure SQLite3 is installed and database directory exists ```bash # Install SQLite3 sudo apt-get install sqlite3 # Ubuntu/Debian brew install sqlite3 # macOS # Create database directory mkdir -p ./.claude/skills/cfn-loop-validation/ # Verify database can be accessed sqlite3 ./.claude/skills/cfn-loop-validation/evidence-chain.db "SELECT 1;" ``` ## Evidence Chain Persistence Validation results are automatically stored in SQLite for audit trails: ```bash # Query validation history sqlite3 ./.claude/skills/cfn-loop-validation/evidence-chain.db \ "SELECT task_id, iteration, confidence, status, timestamp FROM validation_evidence WHERE task_id = 'feature-auth' ORDER BY timestamp DESC;" ``` ## Documentation See `SKILL.md` for comprehensive documentation including: - Validation workflow patterns - Integration with coordinator agents - Evidence chain schema - Redis integration for async validation - Performance targets and telemetry ## Performance Targets - Average execution time: <500ms - Memory footprint: <50MB - Consensus accuracy: >90% - SQLite write latency: <20ms ## Related Skills - **Agent Spawning** - `.claude/skills/agent-spawning/SKILL.md` - **Redis Coordination** - `.claude/skills/redis-coordination/SKILL.md` - **Hook Pipeline** - `.claude/skills/hook-pipeline/SKILL.md` --- **Version:** 2.0.0 **Last Updated:** 2025-10-18 **Maintainer:** Claude Flow Novice Team