Cortivex Node Type Reference

Every node in a Cortivex pipeline is an instance of a node type. Each node type defines a specialized agent role with default behaviors, model assignments, and configuration options.

Node Type Summary Table

Detailed Node Type Reference

1. SecurityScanner

Category: analysis Purpose: Scans source code and dependencies for security vulnerabilities, insecure coding patterns, exposed secrets, and known CVEs in dependencies.

When to use:

• Before merging any PR that touches authentication, authorization, cryptography, or data handling
• As part of nightly or pre-release security audits
• When onboarding a new dependency
• After any code change to security-sensitive modules

Default configuration:

- id: security_scan
  type: SecurityScanner
  config:
    model: claude-sonnet-4-20250514
    scan_depth: standard          # shallow | standard | deep
    check_dependencies: true      # scan package.json / requirements.txt / Cargo.toml etc.
    check_secrets: true           # scan for hardcoded secrets, API keys, tokens
    check_patterns: true          # scan for insecure coding patterns (SQL injection, XSS, etc.)
    severity_threshold: medium    # low | medium | high | critical (only report at or above)
    ignore_paths:                 # paths to exclude from scanning
      - node_modules/
      - vendor/
      - "*.test.*"
    cve_database: latest          # use latest CVE database
    max_files: 500                # limit to avoid excessive cost on huge repos
    output_format: structured     # structured | markdown | sarif

Output:

{
  "vulnerabilities": [
    {
      "severity": "high",
      "type": "sql_injection",
      "file": "src/db/queries.ts",
      "line": 47,
      "description": "User input directly interpolated into SQL query",
      "recommendation": "Use parameterized queries",
      "cwe": "CWE-89"
    }
  ],
  "dependency_issues": [
    {
      "package": "lodash",
      "version": "4.17.15",
      "cve": "CVE-2021-23337",
      "severity": "high",
      "fix_version": "4.17.21"
    }
  ],
  "secrets_found": 0,
  "summary": {
    "total_issues": 3,
    "critical": 0,
    "high": 2,
    "medium": 1,
    "low": 0
  }
}

2. CodeReviewer

Category: analysis Purpose: Performs comprehensive code review, checking for code quality, design patterns, potential bugs, naming conventions, complexity, and adherence to best practices.

When to use:

• On every PR before merging
• When reviewing unfamiliar code
• After large refactors to validate quality
• As part of periodic codebase quality audits

Default configuration:

- id: code_review
  type: CodeReviewer
  config:
    model: claude-sonnet-4-20250514
    review_scope: changed_files   # changed_files | full | directory
    target_path: null             # specific directory to review (when scope is directory)
    check_patterns:               # what to check for
      - error-handling
      - naming-conventions
      - complexity
      - dry-violations
      - dead-code
      - type-safety
      - edge-cases
      - documentation
    max_issues: 50                # cap issues to avoid overwhelming output
    severity_levels: true         # categorize issues by severity
    suggest_fixes: true           # include fix suggestions for each issue
    ignore_patterns:              # file patterns to skip
      - "*.generated.*"
      - "*.min.*"
    complexity_threshold: 15      # cyclomatic complexity warning threshold
    line_length_limit: 120        # flag lines exceeding this length

Output:

{
  "issues": [
    {
      "severity": "warning",
      "category": "complexity",
      "file": "src/utils/parser.ts",
      "line": 23,
      "description": "Function parseInput has cyclomatic complexity of 18 (threshold: 15)",
      "suggestion": "Extract nested conditionals into helper functions"
    }
  ],
  "summary": {
    "files_reviewed": 12,
    "total_issues": 7,
    "errors": 1,
    "warnings": 4,
    "info": 2
  },
  "overall_quality": "good"
}

3. BugHunter

Category: analysis Purpose: Actively searches for bugs, logic errors, race conditions, off-by-one errors, null pointer issues, and other defects that could cause runtime failures.

When to use:

• When investigating reported bugs to find root cause and related issues
• As part of pre-release quality gates
• On complex business logic code
• After major refactors to catch regressions

Default configuration:

- id: bug_hunt
  type: BugHunter
  config:
    model: claude-sonnet-4-20250514
    hunt_scope: changed_files     # changed_files | full | directory
    target_path: null
    bug_categories:               # what types of bugs to hunt
      - null-safety
      - off-by-one
      - race-conditions
      - resource-leaks
      - type-coercion
      - boundary-conditions
      - error-propagation
      - state-mutations
    include_edge_cases: true      # look for edge case inputs that could trigger bugs
    trace_dat