catlog22

Role Analysis Reviewer Agent — Research agent for Claude Code

Research community intermediate

Validates role analysis outputs against role-specific templates and quality standards.

How to install Role Analysis Reviewer Agent

This entry records only its repository, not the path inside it, so there is no exact command to give. Open catlog22/Claude-Code-Workflow and copy the folder into ~/.claude/agents/.

What Role Analysis Reviewer Agent does

Validates role analysis outputs against role-specific templates and quality standards.

Alternatives in Research

Cross Role Reviewer — Compares Decision Digests across role analysis files in a brainstorm session to surface conflicts, gaps, and s 530 ★
Refactor Analyst — Use this agent when you need a thorough code quality analysis and refactoring recommendations for specific fil 358 ★
Verify Figures — A verification agent that performs Stage 1 figure verification — comparing raw figure outputs (PDFs, PNGs from 134 ★

README

Role Analysis Reviewer Agent

Validates role analysis outputs against role-specific templates and quality standards.

Objective

Ensure role analysis documents meet SPEC.md-level precision and completeness.

Input

Role name (e.g., system-architect, product-manager)
Role analysis document path (e.g., .brainstorming/system-architect/analysis.md)
Role template path (e.g., templates/role-templates/system-architect-template.md)

Validation Rules

For system-architect

**MUST have (blocking)**:

Data Model section with at least 1 entity definition
Entity table with fields: name, type, constraint, description
State Machine section with at least 1 lifecycle diagram
State transition table with: from, event, to, side effects
Error Handling Strategy section
Observability Requirements section with at least 3 metrics

**SHOULD have (warning)**:

Data Model with 3-5 entities
Relationships defined between entities
Indexes specified for entities
State Machine with error recovery paths
Circuit Breaker configuration
Health check endpoints defined

**Quality checks**:

All constraints use RFC 2119 keywords
State diagram is valid ASCII art
Metrics have type (Counter/Gauge/Histogram) specified
Error handling has timeout values
No generic placeholders like "[TODO]" or "[TBD]"

Output

JSON validation report:

{
  "role": "system-architect",
  "document": ".brainstorming/system-architect/analysis.md",
  "validation": {
    "must_pass": true/false,
    "should_pass": true/false,
    "checks": {
      "data_model_present": true,
      "entity_count": 3,
      "state_machine_present": true,
      "error_handling_present": true,
      "observability_present": true,
      "metrics_count": 5,
      "rfc_keywords_count": 12
    },
    "issues": [
      {
        "severity": "error",
        "section": "Data Model",
        "message": "Entity 'User' missing 'Relationships' subsection"
      },
      {
        "severity": "warning",
        "section": "Observability",
        "message": "Only 3 metrics defined, recommend at least 5"
      }
    ]
  },
  "score": 8.5,
  "recommendation": "Pass with minor improvements"
}

Execution

Bash({
  command: `ccw cli -p "Validate role analysis document against template.

ROLE: ${roleName}
DOCUMENT: ${analysisDoc}
TEMPLATE: ${roleTemplate}

VALIDATION RULES:
${validationRules}

OUTPUT: JSON validation report with must_pass, should_pass, checks, issues, score, recommendation
" --tool gemini --mode analysis`,
  run_in_background: true
});