workspace-analyzer（工作空间分析器）

Workspace Analyzer

"Scans, analyzes, and reports. The agent decides."

Overview

A self-improving agent needs a clean workspace. This skill analyzes any OpenClaw workspace to:

1. Detect core files dynamically (adapts to workspace changes)
2. Analyze content for issues (bloat, duplicates, broken links)
3. Validate single source of truth (content in multiple places)
4. Report actionable insights for the agent to act upon

Key Principle: The script analyzes → the agent decides → the agent acts.

Installation

# Clone or copy to your skills folder
cp -r workspace-analyzer/ ~/.openclaw/workspace/skills/

Usage

Run Analysis

# Full analysis (default)
python3 skills/workspace-analyzer/scripts/analyzer.py

# Quick scan (structure only, no content analysis)
python3 skills/workspace-analyzer/scripts/analyzer.py --quick

# Specific workspace
python3 skills/workspace-analyzer/scripts/analyzer.py --root /path/to/workspace

# Output to file
python3 skills/workspace-analyzer/scripts/analyzer.py --output report.json

Output Format

{
  "scan_info": {
    "root": "/home/user/.openclaw/workspace",
    "timestamp": "2026-02-21T18:00:00Z",
    "files_scanned": 291
  },
  "core_files_detected": {
    "kai_core": {
      "files": ["SOUL.md", "OPERATING.md", ...],
      "count": 11
    },
    "mission_control": {...},
    "agent_cores": {...},
    "skills": {...}
  },
  "analysis": {
    "SOUL.md": {
      "category": "kai_core",
      "line_count": 450,
      "sections": [...],
      "wiki_links": [...],
      "issues": [...]
    }
  },
  "single_source_validation": {
    "skill_graph": {
      "status": "PASS/FAIL",
      "locations": ["AGENTS.md", "OPERATING.md", "SOUL.md"],
      "recommendation": "Reference all to SUB_CONSCIOUS.md"
    },
    "memory_architecture": {...},
    "image_handling": {...}
  },
  "recommendations": [
    {"action": "FIX_DUPLICATE_CONTENT", "topic": "skill_graph", "files": ["AGENTS.md", "OPERATING.md"], "severity": "CRITICAL"},
    {"action": "REVIEW_BLOAT", "file": "OPERATING.md", "severity": "WARN"},
    {"action": "CHECK_DUPLICATE", "severity": "WARN"},
    {"action": "CHECK_BROKEN_LINK", "severity": "INFO"}
  ],
  "summary": {
    "total_files": 291,
    "total_issues": 17,
    "total_recommendations": 25
  }
}

Single Source of Truth Validation

What It Checks

The analyzer validates that each topic exists in exactly ONE place:

Detection Logic

1. Scan all core files for key section headings
2. Build content map of where each topic appears
3. Flag violations where topic appears in 2+ places
4. Recommend fix by pointing to single source

Example Output

{
  "single_source_validation": {
    "skill_graph": {
      "status": "FAIL",
      "locations": ["AGENTS.md:285", "OPERATING.md:56", "SOUL.md:52"],
      "severity": "CRITICAL",
      "recommendation": "Consolidate to SUB_CONSCIOUS.md, reference from others"
    },
    "memory_architecture": {
      "status": "FAIL", 
      "locations": ["AGENTS.md:157", "OPERATING.md:74", "OPERATING.md:193"],
      "severity": "CRITICAL",
      "recommendation": "Remove duplicate section at OPERATING.md:193"
    }
  }
}

Features

Dynamic Core File Detection

Automatically detects core files based on location patterns:

Category-Specific Thresholds

Bloat thresholds vary by category:

Issue Detection

• BLOAT_WARNING: File exceeds warning threshold
• BLOAT_CRITICAL: File exceeds critical threshold
• ORPHAN_WARNING: File not modified in 30+ days
• DUPLICATE_CONTENT: Same topic in multiple files
• BROKEN_WIKI_LINK: Wiki-link may not match any file

Recommendations

The analyzer generates actionable recommendations:

Fix Instructions for Agents

When duplicate content is detected, follow these steps:

Step 1: Identify the True Source

Determine which file should be the single source:

Step 2: Update Non-Source Files

Replace duplicate content with references:

**Topic:** See [[SUB_CONSCIOUS.md]] for procedures.

Step 3: Verify Bootstrap Injection

Ensure new files are injected at session start:

Check ~/.openclaw/openclaw.json:

{
  "hooks": {
    "internal": {
      "entries": {
        "bootstrap-extra-files": {
          "enabled": true,
          "paths": ["SUB_CONSCIOUS.md", "self-improving/memory.md"]
        }
      }
    }
  }
}

Step 4: Commit Changes

git add -A
git commit -m "Fix duplicate content - consolidate to single source"

Interpretation Guide

How to Use Results

Step 1: Prioritize by Severity

CRITICAL → Review immediately
WARN → Review during next maintenance
INFO → Review during weekly cleanup

Step 2: Check Single Source Validation

FAIL → Fix duplicate content first (most important)
PASS → Move to other issues

Step 3: Understand Context

• BLOAT = Informational, not all need fixing

- Reference docs (skills/*/references/*.md) can be legitimately large - Research logs are consolidations - consider splitting by date - Session logs - archive old ones

• DUPLICATES = Check if intentional

- IDENTITY files = Expected (agent templates) - REVIEW files = May need consolidation - LESSONS files = OK (different skills)

• BROKEN LINKS = Usually false positives

- Links to skills like [[blogwatcher]] ARE valid - Skills have SKILL.md suffix - analyzer doesn't detect this - Only flag if link to core file is truly broken

Step 4: Take Action

1. Don't auto-fix - review first
2. Archive old session logs monthly
3. Split large research logs by topic
4. Keep reference docs as-is (legitimate)
5. Fix duplicate content by consolidating to single source

Integration

With Heartbeat

Add to your HEARTBEAT.md maintenance section:

## 7. Memory + Workspace Maintenance

### Run Workspace Analyzer
python3 skills/workspace-analyzer/scripts/analyzer.py --output /tmp/analysis.json

### Review Single Source Validation
- Check for DUPLICATE_CONTENT issues
- Fix by consolidating to single source

### Review Recommendations
- Check recommendations in output
- Prioritize by severity
- Fix issues manually

Output to Memory

Save analysis results for later review:

python3 skills/workspace-analyzer/scripts/analyzer.py \
  --output memory/$(date +%Y-%m-%d)-workspace-analysis.json

Safety & Security

Read-Only

• Never modifies files
• Only reads and analyzes

No Secrets

• Never reads API keys
• Never accesses credentials
• Only analyzes file metadata and content structure

Safe Output

• Only contains: file paths, sizes, line counts, recommendations
• No sensitive data exposed

Limitations

• No auto-fix: Script reports, agent must decide and act
• Wiki-link false positives: Links to external skills may appear broken
• Date-based duplicate detection: May miss non-date-based duplicates

Examples

Sample Output - Single Source Validation

{
  "single_source_validation": {
    "skill_graph": {
      "status": "FAIL",
      "files": ["AGENTS.md", "OPERATING.md", "SOUL.md"],
      "severity": "CRITICAL",
      "recommendation": "Reference all to SUB_CONSCIOUS.md"
    },
    "memory_architecture": {
      "status": "FAIL",
      "files": ["AGENTS.md", "OPERATING.md"],
      "severity": "CRITICAL", 
      "recommendation": "Remove duplicate in OPERATING.md"
    },
    "message_reactions": {
      "status": "PASS",
      "files": ["SUB_CONSCIOUS.md"],
      "severity": "OK"
    }
  }
}

Sample Output - Recommendations

[
  {
    "action": "FIX_DUPLICATE_CONTENT",
    "topic": "skill_graph",
    "files": ["AGENTS.md:285", "OPERATING.md:56", "SOUL.md:52"],
    "severity": "CRITICAL",
    "recommendation": "Replace with reference to SUB_CONSCIOUS.md"
  },
  {
    "action": "REVIEW_BLOAT",
    "file": "OPERATING.md",
    "category": "kai_core",
    "reason": "503 lines - consider splitting (threshold: 400)",
    "severity": "WARN"
  }
]

Skill Graph

This skill is related to:

• [[mcporter]] - For MCP server analysis
• [[clean-workspace]] - For actual cleanup tasks
• [[qmd]] - For memory organization

Changelog

v1.1.0 (2026-02-24)

• Added single source of truth validation
• Added duplicate content detection (same topics in multiple files)
• Added FIX_DUPLICATE_CONTENT recommendation type
• Added SUB_CONSCIOUS category
• Added fix instructions for agents

v1.0 (2026-02-21)

• Initial release
• Basic file analysis
• Core file detection
• Bloat detection

*Last updated: 2026-02-24*