Agent Skills: Insecure Defaults

Insecure Defaults

Security Notice

AUTHORIZED USE ONLY: These skills are for DEFENSIVE security analysis and authorized research:

• Configuration auditing for owned applications
• Pre-deployment security checks in CI/CD pipelines
• Compliance validation (SOC2, PCI-DSS, HIPAA)
• Security assessment of third-party components
• Educational purposes in controlled environments

NEVER use for:

• Exploiting discovered credentials without authorization
• Accessing systems using found default passwords
• Circumventing security controls
• Any illegal activities

Step 1: Credential Detection

Hardcoded Credential Patterns

Search for credentials embedded directly in source code:

# Password patterns
grep -rnE "(password|passwd|pwd)\s*[=:]\s*['\"][^'\"]{3,}" --include="*.{js,ts,py,java,go,rb,php}" .

# API key patterns
grep -rnE "(api[_-]?key|apikey|api[_-]?secret)\s*[=:]\s*['\"][^'\"]{8,}" --include="*.{js,ts,py,java,go}" .

# Token patterns
grep -rnE "(token|secret|auth)\s*[=:]\s*['\"][A-Za-z0-9+/=]{16,}" --include="*.{js,ts,py,java,go}" .

# AWS credentials
grep -rnE "AKIA[0-9A-Z]{16}" .
grep -rnE "aws_secret_access_key\s*[=:]\s*['\"][^'\"]{20,}" .

# Private keys
grep -rnl "BEGIN (RSA |EC |DSA |OPENSSH )?PRIVATE KEY" .

# Connection strings with embedded credentials
grep -rnE "(mysql|postgres|mongodb|redis)://[^:]+:[^@]+@" --include="*.{js,ts,py,java,go,yml,yaml,json,env}" .

Semgrep Rules for Credential Detection

rules:
  - id: hardcoded-password
    message: >
      Hardcoded password detected. Store passwords in environment
      variables or a secrets manager (AWS Secrets Manager, HashiCorp Vault).
    severity: ERROR
    languages: [javascript, typescript]
    metadata:
      cwe: [CWE-798]
      owasp: [A07:2021]
      confidence: HIGH
      impact: HIGH
      category: security
    patterns:
      - pattern-either:
          - pattern: |
              $VAR = "..."
          - pattern: |
              const $VAR = "..."
          - pattern: |
              let $VAR = "..."
      - metavariable-regex:
          metavariable: $VAR
          regex: (?i)(password|passwd|pwd|secret|credential|api.?key)
      - pattern-not: |
          $VAR = ""
      - pattern-not: |
          $VAR = "changeme"
      - pattern-not: |
          $VAR = "placeholder"

  - id: hardcoded-password-in-config
    message: >
      Password found in configuration file. Use environment variable
      references or encrypted secrets instead.
    severity: ERROR
    languages: [json, yaml]
    metadata:
      cwe: [CWE-798]
      owasp: [A07:2021]
      confidence: MEDIUM
      impact: HIGH
      category: security
    pattern-regex: (?i)(password|secret|api.?key|token)\s*[:=]\s*["']?[a-zA-Z0-9+/=!@#$%^&*]{8,}

Step 2: Fail-Open Detection

What is Fail-Open?

Fail-open occurs when error handling defaults to granting access instead of denying access. This is the opposite of the "fail-secure" principle.

Fail-Open Patterns

// FAIL-OPEN: Error in auth check grants access
try {
  const isAuthorized = await checkAuth(req);
  if (!isAuthorized) return res.status(403).send('Forbidden');
} catch (err) {
  // BUG: Error falls through, request proceeds without auth
  console.log('Auth error:', err);
}
// ... handler continues, even on auth error

// FAIL-SECURE: Error in auth check denies access
try {
  const isAuthorized = await checkAuth(req);
  if (!isAuthorized) return res.status(403).send('Forbidden');
} catch (err) {
  console.error('Auth error:', err);
  return res.status(500).send('Internal error'); // DENY on error
}

Semgrep Rules for Fail-Open

rules:
  - id: fail-open-auth-catch
    message: >
      Authentication/authorization error is caught but execution continues.
      This is a fail-open pattern. Ensure errors in auth checks result
      in access denial, not access grant.
    severity: ERROR
    languages: [javascript, typescript]
    metadata:
      cwe: [CWE-636]
      owasp: [A07:2021]
      confidence: MEDIUM
      impact: HIGH
      category: security
    patterns:
      - pattern: |
          try {
            ...
            $AUTH(...)
            ...
          } catch ($ERR) {
            ...
          }
          ...
          $RESPONSE(...)
      - metavariable-regex:
          metavariable: $AUTH
          regex: (?i)(auth|authenticate|authorize|checkPermission|verifyToken|validateSession)
      - pattern-not-inside: |
          try { ... } catch ($ERR) { ... return ...; }
      - pattern-not-inside: |
          try { ... } catch ($ERR) { ... throw ...; }

  - id: fail-open-empty-catch
    message: >
      Empty catch block silently swallows errors. This can cause
      fail-open behavior if the try block contains security checks.
    severity: WARNING
    languages: [javascript, typescript]
    metadata:
      cwe: [CWE-390]
      confidence: MEDIUM
      impact: MEDIUM
      category: security
    pattern: |
      try { ... } catch ($ERR) { }

Step 3: Insecure Configuration Detection

Common Insecure Defaults

| Category | Insecure Default | Secure Alternative | | --------------- | ---------------------------------- | --------------------------------- | | CORS | origin: '*' | Explicit allowed origins | | Debug | debug: true in production | debug: false | | Binding | 0.0.0.0 (all interfaces) | 127.0.0.1 or specific IP | | TLS | rejectUnauthorized: false | rejectUnauthorized: true | | Cookies | secure: false, httpOnly: false | secure: true, httpOnly: true | | Session | secret: 'keyboard cat' | Random secret from env var | | Crypto | MD5, SHA1, DES, RC4 | SHA-256+, AES-256, ChaCha20 | | Permissions | 0777 file permissions | 0644 or 0600 | | JWT | algorithm: 'none' | algorithm: 'RS256' or 'ES256' | | Rate limit | No rate limiting | Rate limiting enabled | | CSRF | CSRF protection disabled | CSRF protection enabled | | Helmet | No security headers | Helmet middleware enabled |

Semgrep Rules for Insecure Config

rules:
  - id: cors-wildcard-origin
    message: >
      CORS allows all origins ('*'). This permits any website to make
      cross-origin requests. Specify allowed origins explicitly.
    severity: WARNING
    languages: [javascript, typescript]
    metadata:
      cwe: [CWE-942]
      owasp: [A05:2021]
      confidence: HIGH
      impact: MEDIUM
      category: security
    pattern-either:
      - pattern: cors({ origin: '*' })
      - pattern: cors({ origin: true })
      - pattern: |
          $RES.setHeader("Access-Control-Allow-Origin", "*")

  - id: tls-verification-disabled
    message: >
      TLS certificate verification is disabled. This allows
      man-in-the-middle attacks. Remove rejectUnauthorized: false.
    severity: ERROR
    languages: [javascript, typescript]
    metadata:
      cwe: [CWE-295]
      owasp: [A02:2021]
      confidence: HIGH
      impact: HIGH
      category: security
    pattern-either:
      - pattern: rejectUnauthorized: false
      - pattern: process.env.NODE_TLS_REJECT_UNAUTHORIZED = '0'
      - pattern: process.env['NODE_TLS_REJECT_UNAUTHORIZED'] = '0'

  - id: debug-mode-enabled
    message: >
      Debug mode appears to be enabled. Ensure debug mode is
      disabled in production to prevent information disclosure.
    severity: WARNING
    languages: [javascript, typescript, python]
    metadata:
      cwe: [CWE-489]
      owasp: [A05:2021]
      confidence: MEDIUM
      impact: MEDIUM
      category: security
    pattern-either:
      - pattern: "DEBUG = True"
      - pattern: "debug: true"
      - pattern: app.debug = true

  - id: weak-crypto-algorithm
    message: >
      Weak cryptographic algorithm detected. Use SHA-256 or stronger
      for hashing, AES-256 for encryption.
    severity: ERROR
    languages: [javascript, typescript]
    metadata:
      cwe: [CWE-327]
      owasp: [A02:2021]
      confidence: HIGH
      impact: HIGH
      category: security
    patterns:
      - pattern: crypto.createHash($ALGO)
      - metavariable-regex:
          metavariable: $ALGO
          regex: ['"]?(md5|sha1|md4|md2|ripemd)['"]?

Step 4: Environment and Configuration File Scanning

Files to Check

# Environment files
find . -name "*.env*" -not -path "*/node_modules/*"
find . -name ".env.example" -o -name ".env.sample"

# Configuration files
find . -name "config.*" -o -name "settings.*" -o -name "application.*" \
  | grep -E "\.(js|ts|json|yaml|yml|toml|ini|cfg)$"

# Docker configurations
find . -name "Dockerfile*" -o -name "docker-compose*"

# Cloud configurations
find . -name "*.tf" -o -name "*.tfvars" \
  -o -name "serverless.*" -o -name "*.cloudformation.*"

Configuration Audit Checklist

## Configuration Security Audit

### Credentials

- [ ] No hardcoded passwords in source code
- [ ] No API keys in source code
- [ ] No private keys committed
- [ ] No database credentials in config files
- [ ] .env files are in .gitignore
- [ ] .env.example has placeholder values only

### Network

- [ ] Not binding to 0.0.0.0 in production
- [ ] TLS/SSL enabled for all connections
- [ ] Certificate verification enabled
- [ ] CORS restricted to known origins

### Authentication

- [ ] Session secret is random and from env var
- [ ] JWT algorithm is not 'none'
- [ ] Password hashing uses bcrypt/argon2 (not MD5/SHA1)
- [ ] Cookie settings: secure, httpOnly, sameSite

### Error Handling

- [ ] Auth errors deny access (fail-secure)
- [ ] No empty catch blocks in security paths
- [ ] Stack traces not exposed in production
- [ ] Default error handler returns 500, not 200

### Infrastructure

- [ ] Debug mode disabled in production
- [ ] File permissions are restrictive
- [ ] Default admin accounts disabled/changed
- [ ] Rate limiting enabled
- [ ] Security headers configured (Helmet/equivalent)

Step 5: Reporting

Insecure Defaults Report

## Insecure Defaults Assessment

**Date**: YYYY-MM-DD
**Scope**: [project/repository]

### Summary

| Category              | Findings | Critical | High  | Medium | Low   |
| --------------------- | -------- | -------- | ----- | ------ | ----- |
| Hardcoded credentials | X        | X        | X     | X      | X     |
| Fail-open patterns    | X        | X        | X     | X      | X     |
| Insecure config       | X        | X        | X     | X      | X     |
| Weak crypto           | X        | X        | X     | X      | X     |
| **Total**             | **X**    | **X**    | **X** | **X**  | **X** |

### Critical Findings

[Detailed findings with remediation]

### Recommended Secure Defaults

[Project-specific secure configuration template]

Default Credential Databases

When auditing for default credentials, check against known defaults:

| Product/Service | Default Username | Default Password | | --------------- | ---------------- | -------------------- | | PostgreSQL | postgres | postgres | | MySQL | root | (empty) | | MongoDB | (none) | (none - no auth) | | Redis | (none) | (none - no auth) | | RabbitMQ | guest | guest | | Elasticsearch | elastic | changeme | | Jenkins | admin | (from initial setup) | | Grafana | admin | admin |

Rule: If your application ships with any of these defaults, flag as CRITICAL.

Related Skills

• static-analysis - CodeQL and Semgrep scanning
• variant-analysis - Finding variants of insecure patterns
• semgrep-rule-creator - Custom detection rules
• differential-review - Detecting new insecure defaults in diffs
• security-architect - STRIDE threat modeling
• auth-security-expert - Authentication best practices

Agent Integration

• security-architect (primary): Security configuration assessment
• code-reviewer (primary): Configuration review in PRs
• penetration-tester (secondary): Credential verification
• devops (secondary): Infrastructure configuration auditing

Iron Laws

1. ALWAYS treat any hardcoded credential or token as a CRITICAL finding regardless of context — "test" or "example" credentials are regularly copied into production and must be remediated immediately.
2. NEVER use rejectUnauthorized: false or equivalent TLS bypass even temporarily — TLS verification disabled "temporarily" frequently ships to production; there is no safe exception.
3. ALWAYS scan configuration files (.env, YAML, JSON, TOML) in addition to source code — credentials are more commonly embedded in config files than in application code.
4. NEVER report a "no findings" result without running automated scanning tools (Semgrep, Trufflehog, git-secrets) — manual review alone misses encoded, encoded, or obfuscated credentials.
5. ALWAYS verify that security checks fail-secure (deny access) by default — fail-open configurations (where auth errors allow access) are the most dangerous class of insecure default.

Anti-Patterns

| Anti-Pattern | Why It Fails | Correct Approach | | ---------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------ | ------------------------------------------------------------------------------------------------- | | Treating "example" credentials as low-severity | Example values are copy-pasted into production; JIRA/Slack contain real production secrets "temporarily" | Flag all hardcoded credentials CRITICAL; rotate immediately if in git history | | Manual-only credential scan | Human eyes miss base64-encoded secrets, multi-line keys, and rotation-obscured patterns | Run Trufflehog + Semgrep + git-secrets on every commit; manual review supplements, never replaces | | rejectUnauthorized: false in test code | Test configs leak into production via copy-paste; TLS bypass is invisible to runtime monitoring | Use NODE_EXTRA_CA_CERTS for custom CAs; never disable peer verification | | Not checking configuration files | 60%+ of credential leaks originate in config files, not source code | Explicitly scan .env, *.yaml, *.json, *.toml, docker-compose.yml | | Assuming framework defaults are secure | Flask debug=True, Django DEBUG=True, Spring Boot actuators — framework defaults are developer-optimized, not production-secure | Enumerate framework-specific insecure defaults; validate each for production environment |

Memory Protocol (MANDATORY)

Before starting: Read .claude/context/memory/learnings.md

After completing:

• New pattern -> .claude/context/memory/learnings.md
• Issue found -> .claude/context/memory/issues.md
• Decision made -> .claude/context/memory/decisions.md

ASSUME INTERRUPTION: If it's not in memory, it didn't happen.