Claude Code Plugins

Community-maintained marketplace

Feedback

project-analyzer

@fractary/claude-plugins
2
0

Analyzes Claude Code project structure detecting architectural patterns and anti-patterns by executing detection scripts

Install Skill

1Download skill
2Enable skills in Claude

Open claude.ai/settings/capabilities and find the "Skills" section

3Upload to Claude

Click "Upload skill" and select the downloaded ZIP file

Note: Please verify skill by going through its instructions before using it.

SKILL.md

name project-analyzer
description Analyzes Claude Code project structure detecting architectural patterns and anti-patterns by executing detection scripts
model claude-haiku-4-5

Project Analyzer Skill

You analyze Claude Code projects by executing detection scripts that check for architectural patterns and anti-patterns.

All detection is done by bash scripts in the scripts/ directory. You execute the scripts and return their output.

1. ALL detection is done by executing bash scripts 2. DO NOT analyze files directly - run the scripts 3. DO NOT fabricate or hallucinate results 4. Return script output as the authoritative result 5. Read-only analysis - never modify project files

run-full-audit

Run all detection scripts and return comprehensive audit results.

Input:

  • project_path: Path to Claude Code project root

Execute:

bash plugins/faber-agent/skills/project-analyzer/scripts/run-all-detections.sh {project_path}

Output: JSON from script containing:

  • summary.compliance_score: Overall compliance percentage
  • summary.total_violations: Count of all violations
  • summary.by_severity: Breakdown by error/warning/info
  • detections.*: Results from each detection script
  • structure: Project structure info

detect-project-specific-director (NEW v2.0 - ERROR SEVERITY)

Detect project-specific director skills - these are anti-patterns and BLOCK workflows.

Input:

  • project_path: Path to Claude Code project root

Execute:

bash plugins/faber-agent/skills/project-analyzer/scripts/detect-project-specific-director.sh {project_path}

Detection Logic:

  1. Search for skill files matching *-director/SKILL.md pattern
  2. Exclude core FABER director (faber-director)
  3. Search for commands with -direct suffix
  4. Check for pattern expansion logic in skills

Output: JSON with:

{
  "status": "error|clean",
  "rule_id": "ARC-006",
  "severity": "error",
  "message": "Project-specific director detected - use core faber-director instead",
  "violations": [
    {
      "file": "skills/{project}-director/SKILL.md",
      "line": null,
      "description": "Project-specific director skill",
      "migration_proposal": {
        "action": "delete",
        "replacement": "Use core faber-director with workflow config",
        "workflow_config_example": "..."
      }
    }
  ]
}

detect-project-specific-manager (NEW v2.0 - ERROR SEVERITY)

Detect project-specific manager agents - these are anti-patterns and BLOCK workflows.

Input:

  • project_path: Path to Claude Code project root

Execute:

bash plugins/faber-agent/skills/project-analyzer/scripts/detect-project-specific-manager.sh {project_path}

Detection Logic:

  1. Search for agent files matching *-manager.md pattern
  2. Exclude core FABER manager (faber-manager)
  3. Check for orchestration/workflow logic in agents
  4. Identify manager agents by content patterns

Output: JSON with:

{
  "status": "error|clean",
  "rule_id": "ARC-007",
  "severity": "error",
  "message": "Project-specific manager detected - use core faber-manager with workflow config",
  "violations": [
    {
      "file": "agents/{project}-manager.md",
      "line": null,
      "description": "Project-specific manager agent",
      "migration_proposal": {
        "action": "delete",
        "replacement": "Use core faber-manager with workflow config",
        "workflow_config": {
          "id": "{project}-workflow",
          "phases": "..."
        },
        "invocation": "/faber run <id> --workflow {project}-workflow"
      }
    }
  ]
}

inspect-structure

Scan project and collect structural information.

Input:

  • project_path: Path to Claude Code project root

Execute:

bash plugins/faber-agent/skills/project-analyzer/scripts/inspect-structure.sh {project_path}

Output: JSON with agents, skills, commands counts and locations.

detect-manager-as-skill

Detect Manager-as-Skill anti-pattern.

Input:

  • project_path: Path to Claude Code project root

Execute:

bash plugins/faber-agent/skills/project-analyzer/scripts/detect-manager-as-skill.sh {project_path}

Output: JSON with detection results.

detect-director-as-agent

Detect Director-as-Agent anti-pattern.

Input:

  • project_path: Path to Claude Code project root

Execute:

bash plugins/faber-agent/skills/project-analyzer/scripts/detect-director-as-agent.sh {project_path}

Output: JSON with detection results.

detect-workflow-logging

Detect missing workflow event logging in manager agents (AGT-005).

Input:

  • project_path: Path to Claude Code project root

Execute:

bash plugins/faber-agent/skills/project-analyzer/scripts/detect-workflow-logging.sh {project_path}

Output: JSON with detection results.

detect-direct-skill-commands

Detect commands that invoke skills directly instead of routing through manager (CMD-004).

Input:

  • project_path: Path to Claude Code project root

Execute:

bash plugins/faber-agent/skills/project-analyzer/scripts/detect-direct-skill-commands.sh {project_path}

Output: JSON with detection results.

detect-director-patterns

Detect missing director argument patterns (ARC-004).

Input:

  • project_path: Path to Claude Code project root

Execute:

bash plugins/faber-agent/skills/project-analyzer/scripts/detect-director-patterns.sh {project_path}

Output: JSON with detection results.

calculate-context-load

Calculate context token load and optimization opportunities.

Input:

  • project_path: Path to Claude Code project root

Execute:

bash plugins/faber-agent/skills/project-analyzer/scripts/calculate-context-load.sh {project_path}

Output: JSON with context load analysis.

 Upon completion, output: ``` ✅ COMPLETED: Project Analyzer Project: {project_path} Results returned to agent ``` **Project not found:** ```json {"status": "error", "error": "project_not_found", "message": "Directory does not exist"} ```

Not a Claude Code project:

{"status": "error", "error": "invalid_project", "message": ".claude/ directory not found"}

Script failed:

{"status": "error", "error": "script_failed", "message": "Script execution error"}