claude-flow-novice 2.14.9 → 2.14.11

package/claude-assets/skills/cfn-task-decomposition/task-decomposer.sh

@@ -0,0 +1,279 @@
+#!/bin/bash
+# Task Decomposition for Complex Operations
+# Breaks down complex tasks into smaller, tool-budget-efficient subtasks
+
+set -euo pipefail
+
+# Configuration
+TASK_ID="${1}"
+TASK_DESCRIPTION="${2}"
+TOOL_BUDGET="${3:-10}"  # Default 10 tool budget per agent
+COMPLEXITY_THRESHOLD="${4:-medium}"  # low, medium, high
+
+# Colors for output
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+BLUE='\033[0;34m'
+NC='\033[0m' # No Color
+
+echo -e "${BLUE}🔧 CFN Loop Task Decomposition${NC}"
+echo "Task ID: $TASK_ID"
+echo "Tool Budget: $TOOL_BUDGET per agent"
+echo "Complexity: $COMPLEXITY_THRESHOLD"
+echo ""
+
+# Function to analyze task complexity
+analyze_task_complexity() {
+    local description="$1"
+    local complexity_score=0
+
+    # File operation indicators
+    if echo "$description" | grep -qiE "(fix|create|modify|update|delete|write).*file"; then
+        ((complexity_score += 3))
+    fi
+
+    # Directory exploration indicators
+    if echo "$description" | grep -qiE "(explore|find|search|scan|directory|folder)"; then
+        ((complexity_score += 2))
+    fi
+
+    # Multiple component indicators
+    if echo "$description" | grep -qiE "(multiple|batch|all|every|comprehensive)"; then
+        ((complexity_score += 4))
+    fi
+
+    # TypeScript/Complex file indicators
+    if echo "$description" | grep -qiE "(typescript|tsx|interface|type|complex)"; then
+        ((complexity_score += 2))
+    fi
+
+    # Dependency analysis indicators
+    if echo "$description" | grep -qiE "(dependency|import|require|module)"; then
+        ((complexity_score += 2))
+    fi
+
+    echo "$complexity_score"
+}
+
+# Function to decompose TypeScript error fixing tasks
+decompose_typescript_task() {
+    local task_id="$1"
+    local description="$2"
+    local output_file=".claude/skills/cfn-task-decomposition/${task_id}-subtasks.json"
+
+    echo -e "${YELLOW}Decomposing TypeScript task...${NC}"
+
+    # Create subtasks
+    cat > "$output_file" << EOF
+{
+  "task_id": "$task_id",
+  "original_task": "$description",
+  "decomposition_strategy": "typescript-error-fixing",
+  "subtasks": [
+    {
+      "subtask_id": "${task_id}-recon",
+      "title": "Directory Reconnaissance",
+      "description": "Explore project structure and identify TypeScript files needing fixes",
+      "agent_type": "researcher",
+      "tool_budget": 5,
+      "expected_tools": ["find", "grep", "Read", "Glob"],
+      "deliverables": ["file-list.txt", "error-classification.json"],
+      "estimated_duration": "5-10 minutes"
+    },
+    {
+      "subtask_id": "${task_id}-pattern",
+      "title": "Error Pattern Analysis",
+      "description": "Analyze TypeScript errors and identify common patterns",
+      "agent_type": "analyst",
+      "tool_budget": 8,
+      "expected_tools": ["Read", "Grep", "Bash"],
+      "deliverables": ["error-patterns.json", "fix-strategy.md"],
+      "estimated_duration": "10-15 minutes"
+    },
+    {
+      "subtask_id": "${task_id}-fix",
+      "title": "File-by-File Fixes",
+      "description": "Fix TypeScript errors in identified files using patterns",
+      "agent_type": "backend-developer",
+      "tool_budget": 15,
+      "expected_tools": ["Read", "Edit", "Write", "Bash"],
+      "deliverables": ["fixed-files/", "fix-summary.json"],
+      "estimated_duration": "20-30 minutes",
+      "batch_processing": {
+        "max_files_per_batch": 3,
+        "batches_total": 5
+      }
+    },
+    {
+      "subtask_id": "${task_id}-validate",
+      "title": "Validation and Testing",
+      "description": "Validate fixes and run TypeScript compilation",
+      "agent_type": "tester",
+      "tool_budget": 10,
+      "expected_tools": ["Bash", "Read", "Write"],
+      "deliverables": ["validation-report.json", "compilation-results.txt"],
+      "estimated_duration": "10-15 minutes"
+    }
+  ],
+  "coordination": {
+    "dependency_order": ["recon", "pattern", "fix", "validate"],
+    "redis_context_keys": {
+      "recon": "${task_id}:recon:results",
+      "pattern": "${task_id}:pattern:results",
+      "fix": "${task_id}:fix:results",
+      "validate": "${task_id}:validate:results"
+    }
+  }
+}
+EOF
+
+    echo -e "${GREEN}✓ Task decomposition created: $output_file${NC}"
+    echo "$output_file"
+}
+
+# Function to decompose file exploration tasks
+decompose_exploration_task() {
+    local task_id="$1"
+    local description="$2"
+    local output_file=".claude/skills/cfn-task-decomposition/${task_id}-subtasks.json"
+
+    echo -e "${YELLOW}Decomposing exploration task...${NC}"
+
+    cat > "$output_file" << EOF
+{
+  "task_id": "$task_id",
+  "original_task": "$description",
+  "decomposition_strategy": "directory-exploration",
+  "subtasks": [
+    {
+      "subtask_id": "${task_id}-scan",
+      "title": "Targeted Directory Scan",
+      "description": "Scan specific directories for target files using ripgrep",
+      "agent_type": "researcher",
+      "tool_budget": 6,
+      "expected_tools": ["Bash", "rg", "Glob", "Read"],
+      "deliverables": ["target-files.txt", "directory-map.json"],
+      "estimated_duration": "5-8 minutes"
+    },
+    {
+      "subtask_id": "${task_id}-analyze",
+      "title": "File Content Analysis",
+      "description": "Analyze found files to understand structure and patterns",
+      "agent_type": "analyst",
+      "tool_budget": 8,
+      "expected_tools": ["Read", "Grep", "Write"],
+      "deliverables": ["file-analysis.json", "content-summary.md"],
+      "estimated_duration": "8-12 minutes"
+    }
+  ]
+}
+EOF
+
+    echo -e "${GREEN}✓ Task decomposition created: $output_file${NC}"
+    echo "$output_file"
+}
+
+# Function to create task-specific agent prompts
+create_agent_prompts() {
+    local subtasks_file="$1"
+    local prompts_dir=".claude/skills/cfn-task-decomposition/prompts"
+
+    mkdir -p "$prompts_dir"
+
+    # Extract subtask information
+    local task_id=$(jq -r '.task_id' "$subtasks_file")
+    local strategy=$(jq -r '.decomposition_strategy' "$subtasks_file")
+
+    # Generate prompts for each subtask
+    jq -r '.subtasks[] | @base64' "$subtasks_file" | while read -r subtask_b64; do
+        local subtask=$(echo "$subtask_b64" | base64 -d)
+        local subtask_id=$(echo "$subtask" | jq -r '.subtask_id')
+        local title=$(echo "$subtask" | jq -r '.title')
+        local description=$(echo "$subtask" | jq -r '.description')
+        local tool_budget=$(echo "$subtask" | jq -r '.tool_budget')
+        local deliverables=$(echo "$subtask" | jq -r '.deliverables[]')
+
+        cat > "$prompts_dir/${subtask_id}-prompt.md" << EOF
+# Task: $title
+
+## Context
+You are working on subtask "${subtask_id}" of the larger task "${task_id}".
+
+## Your Mission
+$description
+
+## Tool Budget Optimization
+You have **$tool_budget tool uses** available. Use them efficiently:
+
+### Recommended Tool Sequence:
+1. **Exploration tools** (Read, Glob, Bash) - 2-3 uses
+2. **Analysis tools** (Grep, Read) - 2-3 uses
+3. **Implementation tools** (Edit, Write) - 3-5 uses
+4. **Validation tools** (Bash, Read) - 1-2 uses
+
+### Efficiency Tips:
+- **Batch operations**: Read multiple files in one tool use when possible
+- **Targeted searches**: Use specific patterns instead of broad scans
+- **Early validation**: Check results after each major step
+- **Combine tool uses**: Use compound commands to reduce tool count
+
+## Expected Deliverables
+$deliverables
+
+## Success Criteria
+- All deliverables created with high quality
+- Tool budget not exceeded
+- Results passed to next subtask via Redis
+
+## Redis Context
+- Store results in: \`${task_id}:${subtask_id}:results\`
+- Use confidence scoring based on deliverable completion
+- Signal completion via Redis LPUSH to \`${task_id}:${subtask_id}:done\`
+EOF
+
+        echo -e "${GREEN}✓ Created prompt for $subtask_id${NC}"
+    done
+}
+
+# Main execution
+main() {
+    local complexity_score=$(analyze_task_complexity "$TASK_DESCRIPTION")
+    echo -e "${YELLOW}Task Complexity Score: $complexity_score${NC}"
+
+    local decomposition_file=""
+
+    # Determine decomposition strategy based on task type
+    if echo "$TASK_DESCRIPTION" | grep -qiE "(typescript|tsx|ts27|ts23|error|fix)"; then
+        decomposition_file=$(decompose_typescript_task "$TASK_ID" "$TASK_DESCRIPTION")
+    elif echo "$TASK_DESCRIPTION" | grep -qiE "(explore|find|search|scan)"; then
+        decomposition_file=$(decompose_exploration_task "$TASK_ID" "$TASK_DESCRIPTION")
+    else
+        echo -e "${YELLOW}Using general task decomposition...${NC}"
+        decomposition_file=$(decompose_typescript_task "$TASK_ID" "$TASK_DESCRIPTION")
+    fi
+
+    # Create agent prompts
+    if [ -n "$decomposition_file" ] && [ -f "$decomposition_file" ]; then
+        create_agent_prompts "$decomposition_file"
+
+        echo -e "\n${GREEN}=== Task Decomposition Complete ===${NC}"
+        echo -e "📄 ${BLUE}Decomposition file:${NC} $decomposition_file"
+        echo -e "🤖 ${BLUE}Agent prompts:${NC} .claude/skills/cfn-task-decomposition/prompts/"
+        echo -e "💡 ${BLUE}Strategy:${NC} Break complex task into tool-budget-efficient subtasks"
+
+        # Store decomposition info in Redis for coordinator
+        if command -v redis-cli >/dev/null 2>&1; then
+            redis-cli set "${TASK_ID}:decomposition" "$(cat "$decomposition_file")" >/dev/null
+            echo -e "📦 ${BLUE}Redis context stored:${NC} ${TASK_ID}:decomposition"
+        fi
+
+        return 0
+    else
+        echo -e "${RED}❌ Task decomposition failed${NC}"
+        return 1
+    fi
+}
+
+# Execute main function
+main "$@"

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "claude-flow-novice",
-  "version": "2.14.9",
+  "version": "2.14.11",
   "description": "AI agent orchestration framework with namespace-isolated skills, agents, and CFN Loop validation. Safe installation with ~0.01% collision risk.",
   "main": "dist/index.js",
   "type": "module",