What is project-evaluation?

The Project Evaluation skill provides a sophisticated framework for assessing software project health using multi-agent coordination (Hive-Mind) and persistent memory (AgentDB). It automates the analysis of system architecture, tracks critical performance metrics, and utilizes Goal-Oriented Action Planning (GOAP) to generate strategic refactoring roadmaps, ensuring long-term code quality and project transparency through neural pattern analysis.

When should I use project-evaluation?

project-evaluation is useful in the following scenarios: • Architecture Health Assessment: Evaluate complex system designs to identify structural weaknesses, domain separation issues, and technical debt. • Automated Technical Reporting: Generate data-driven status reports including test coverage, bundle sizes, and build performance metrics for stakeholders. • Strategic Refactoring Planning: Utilize GOAP (Goal-Oriented Action Planning) to create actionable, step-by-step sequences for improving project grades and resolving critical issues. • Cross-Session Knowledge Persistence: Use AgentDB to maintain a 'memory' of previous evaluations, allowing the AI to track project evolution and recurring patterns over time.

name	"Project Evaluation"
description	"Comprehensive project status evaluation using hive-mind coordination, GOAP planning, neural analysis, and AgentDB memory. Use when assessing architecture health, planning refactoring, or generating status reports."

Project Evaluation Skill

What This Skill Does

Orchestrates comprehensive project evaluation using all Claude Flow systems:

Hive-Mind: Collective intelligence coordination
AgentDB: Persistent memory across sessions
Neural Training: Pattern learning from evaluations
GOAP Planning: Action planning for improvements
Skill Creation: Document learnings as reusable skills

Prerequisites

Claude Flow v2.0+ (npx claude-flow@alpha)
Initialized hive-mind (npx claude-flow hive-mind init)
Project with prior session memory

Quick Start

# 1. Retrieve prior session state
npx claude-flow memory retrieve --namespace hive-mind --key "session/*"

# 2. Initialize evaluation swarm
npx claude-flow swarm init --topology hierarchical --agents 5

# 3. Spawn evaluation agents
# Use Claude Code Task tool:
Task("Architecture Agent", "Evaluate architecture health...", "system-architect")
Task("GOAP Agent", "Generate improvement plan...", "code-goal-planner")

Evaluation Framework

Phase 1: Memory Retrieval

// Retrieve prior session state
const session = await memory.retrieve('session/*/completed', 'hive-mind');
const worldState = await memory.retrieve('goap/world-state/final', 'goap');

Phase 2: Agent Spawning

// Spawn evaluation agents via Task tool
[Parallel]:
  Task("system-architect", "Evaluate architecture against assessment...")
  Task("code-goal-planner", "Generate GOAP plan for Grade A...")
  Task("tester", "Analyze test coverage and quality...")

Phase 3: Neural Analysis

// Train on evaluation patterns
await neuralTrain({
  pattern_type: "coordination",
  training_data: { metrics: ["architecture", "testing", "performance"] }
});

Phase 4: Results Storage

// Store in AgentDB for persistence
await memory.store('evaluation/architecture-grade', results, 'agentdb');
await memory.store('evaluation/goap-plan', plan, 'goap');

Evaluation Metrics

Category	Metrics
Architecture	Grade, critical issues, domain separation
Testing	File count, coverage %, pass rate
Performance	Bundle size, build time, store LOC
Tech Debt	Deprecated code, uncommitted changes

Output Format

{
  "evaluation": {
    "previousGrade": "B-",
    "currentGrade": "B+",
    "criticalIssuesResolved": 3,
    "criticalIssuesRemaining": 0
  },
  "goapPlan": {
    "totalCost": 13,
    "actions": ["commit", "test", "delete", "optimize"],
    "successProbability": "85%"
  },
  "metrics": {
    "storeLOC": 3678,
    "testFiles": 67,
    "uncommittedFiles": 19
  }
}

Integration with Other Skills

store-migration-workflow - For refactoring execution
hive-mind-advanced - For collective coordination
agentdb-memory-patterns - For persistence
goap-planning - For action sequencing

Best Practices

Always retrieve prior session state before evaluation
Store all findings in AgentDB for cross-session persistence
Train neural patterns on successful evaluations
Generate GOAP plans for actionable next steps
Create skills from recurring evaluation patterns

Created: 2025-12-03 Version: 1.0.0 Category: Project Management

project-evaluation

When & Why to Use This Skill

Use Cases