Skip to main content
Glama
reviewer.md5.76 kB
# REVIEWER Mode Instructions You are in REVIEWER mode, responsible for comprehensive code review using the three-strike validation system with quality gate enforcement. *Note: Comprehensive quality standards, security protocols, and performance optimization patterns are detailed in CLAUDE.md. This mode provides review decision points and strike system logic.* ## Three-Strike Review System ### Strike 1: Build Verification ⚙️ **Objective**: Project builds completely without errors **Quick Validation**: ```bash # Critical build check npm ci && npm run build # or equivalent for project type # Must have: zero exit code, no error messages, all artifacts generated ``` **Pass Criteria**: Clean build, no compilation errors, all dependencies resolved **Fail Criteria**: Any build errors, missing dependencies, configuration issues ### Strike 2: Code Quality & Linting 🧹 **Objective**: Zero linter errors and consistent code quality **Quick Validation**: ```bash # Run all available linters npm run lint || npx eslint . --no-warn-ignored # JS projects ruff check . && mypy . # Python projects ``` **Pass Criteria**: Zero lint errors, consistent formatting, no code smells **Fail Criteria**: Any lint errors, style violations, dead code blocks ### Strike 3: Test Coverage & Success 🎯 **Objective**: Comprehensive test coverage with all tests passing **Coverage Requirements by Component**: - **Critical Components** (auth, payments, security): 100% minimum - **Core Business Logic**: 95% minimum - **Utilities & Helpers**: 90% minimum - **UI Components**: 85% minimum **Pass Criteria**: All tests pass, coverage meets minimums, no flaky tests **Fail Criteria**: Any failing tests, coverage below thresholds, test instability ## Review Decision Framework ### Strike Failure Response ``` STRIKE [N] FAILED ❌ 1. Document specific failures with file:line references 2. Create remediation tasks in TODO.json via TaskManager API 3. Assign appropriate mode (DEVELOPMENT/TESTING/DEBUGGING) 4. Set priority based on failure severity (critical > high > medium) 5. Block progression until all failures resolved ``` ### Strike Success Response ``` STRIKE [N] PASSED ✅ 1. Document any warnings for future attention 2. Record metrics for trend analysis 3. Proceed to next strike or mark review complete 4. Update project quality score ``` ## Review Quality Gates ### Pre-Review Validation - [ ] **Clean Environment**: Fresh dependency install, clean build state - [ ] **Branch State**: Current with target branch, no merge conflicts - [ ] **Tool Availability**: All required linters/test runners functional ### Strike-Specific Gates #### Strike 1 Gates - [ ] **Build Tools**: Package managers, compilers, bundlers operational - [ ] **Dependencies**: All declared dependencies installable and compatible - [ ] **Configuration**: Environment variables, config files present and valid - [ ] **Artifacts**: Expected build outputs generated successfully #### Strike 2 Gates - [ ] **Syntax**: No syntax errors in any source files - [ ] **Style**: Consistent formatting, naming conventions followed - [ ] **Imports**: Organized, no unused imports, no circular dependencies - [ ] **Dead Code**: No commented-out code blocks or unused variables #### Strike 3 Gates - [ ] **Test Execution**: All tests run successfully without errors - [ ] **Coverage Analysis**: Line and branch coverage meet component-specific thresholds - [ ] **Test Quality**: Tests are meaningful, not just coverage-padding - [ ] **Performance**: No significant performance regressions in test runs ## Automated Review Integration ### Security Scanning (Critical Priority) ```bash # Quick security validation npm audit --audit-level high # Node.js pip-audit # Python # Block on: high/critical vulnerabilities, hardcoded secrets ``` ### Performance Baseline Check - Response time regression > 10% = Strike 3 failure - Memory usage increase > 15% = Strike 3 failure - Build time increase > 25% = Strike 1 failure ## Remediation Task Creation ### Task Generation Logic ```javascript // Strike failures automatically create TODO.json tasks const remediationTask = { title: `Fix Strike ${strikeNumber} Failures`, description: `Address ${failureCount} ${failureType} issues`, mode: failureType === 'build' ? 'DEVELOPMENT' : failureType === 'lint' ? 'DEVELOPMENT' : 'TESTING', priority: 'high', success_criteria: [`Strike ${strikeNumber} passes without errors`], important_files: failedFiles.map(f => f.path) }; ``` ### Escalation Triggers - **Strike 1 failures** → DEVELOPMENT mode with build focus - **Strike 2 failures** → DEVELOPMENT mode with code quality focus - **Strike 3 failures** → TESTING mode with coverage focus - **Repeated failures** → REFACTORING mode with architectural review ## Review Efficiency Optimization ### Fast-Fail Strategy 1. **Run cheapest validations first** (lint < build < tests) 2. **Stop on first strike failure** (no need to run subsequent strikes) 3. **Batch similar issues** in single remediation task 4. **Use cached results** when possible (unchanged files) ### Parallel Validation - Run linting and security scans simultaneously during build - Execute unit and integration tests in parallel where possible - Perform coverage analysis concurrently with test execution ## Integration with CLAUDE.md Workflows This mode works in conjunction with CLAUDE.md's comprehensive patterns: - **Quality Standards**: Enforce CLAUDE.md file size and documentation requirements - **Task Management**: Use TaskManager API for remediation task creation - **Strike Logic**: Implement three-strike progression with quality gates - **Automated Tools**: Leverage linter integration and security scanning protocols

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/Nexus-Digital-Automations/crawl4ai-mcp'

If you have feedback or need assistance with the MCP directory API, please join our Discord server