Critical Interval Security Checker Agent Skill

Critical Interval Security Checker

Analyze code to identify security-critical time intervals and timing vulnerabilities that could compromise security.

Workflow

1. Identify Security-Critical Operations

Look for code that handles:

• Authentication: Login, logout, token generation
• Session management: Session creation, validation, expiration
• Password operations: Reset tokens, change requests
• Rate limiting: Login attempts, API calls, password resets
• Token operations: JWT, OAuth tokens, API keys
• MFA: OTP codes, push notifications, backup codes

2. Check for Common Vulnerabilities

See vulnerability_patterns.md for detailed patterns.

Critical checks:

• Missing expiration validation
• Excessive timeout values
• No rate limiting on sensitive endpoints
• Client-side only expiration checks
• Race conditions in expiration logic
• Hardcoded timeout values
• Inconsistent timeout enforcement
• Timezone handling issues
• Timing attack vulnerabilities

3. Use Automated Checker

Run the automated checker script:

# Check single file
python scripts/check_intervals.py path/to/file.py

# Check entire directory
python scripts/check_intervals.py src/

# Specify language
python scripts/check_intervals.py src/ --language python

The script detects:

• 🔴 Critical: JWT decode without verification, timing attacks
• 🟠 High: Missing expiration checks, no rate limiting
• 🟡 Medium: Excessive timeouts, inconsistent enforcement
• 🔵 Low: Hardcoded timeouts, magic numbers

4. Manual Code Review

For each security-critical operation, verify:

Expiration is set:

# Good: Expiration set
token_expiry = datetime.utcnow() + timedelta(hours=1)

Expiration is checked:

# Good: Expiration validated
if datetime.utcnow() > token_expiry:
    raise TokenExpiredError()

Timeout is reasonable:

# Good: 1-hour reset token
RESET_TOKEN_EXPIRY = timedelta(hours=1)

# Bad: 7-day reset token (too long!)
RESET_TOKEN_EXPIRY = timedelta(days=7)

Rate limiting exists:

# Good: Rate limited
@limiter.limit("5 per 15 minutes")
def login():
    pass

Server-side enforcement:

# Good: Server validates expiration
decoded = jwt.decode(token, SECRET_KEY)  # Checks exp

# Bad: Client-side only
decoded = jwt.decode(token, options={"verify_signature": False})

5. Verify Against Standards

Consult time_intervals.md for recommended values:

Common intervals:

• JWT access token: 5-15 minutes
• JWT refresh token: 7-30 days
• Session timeout: 15-30 minutes (idle), 8-12 hours (absolute)
• Password reset token: 15-60 minutes
• Login rate limit: 5 attempts per 15 minutes
• TOTP code: 30 seconds
• SMS/Email OTP: 5-10 minutes

Check if intervals match use case:

• High-security apps need shorter timeouts
• User-facing apps balance security and UX
• API tokens can be shorter-lived with refresh tokens

6. Document Findings

For each issue found, document:

Issue: Missing expiration check on password reset token
Location: auth/reset_password.py:45
Severity: High
Current: Token created with expiry but never validated
Recommendation: Add expiration check before using token
Code fix:
  if datetime.utcnow() > token.expires_at:
      raise TokenExpiredError("Reset token expired")

7. Propose Fixes

Fix missing expiration checks:

# Before
def validate_token(token):
    decoded = jwt.decode(token, SECRET_KEY, options={"verify_signature": False})
    return decoded['user_id']

# After
def validate_token(token):
    try:
        decoded = jwt.decode(token, SECRET_KEY)  # Verifies exp automatically
        return decoded['user_id']
    except jwt.ExpiredSignatureError:
        raise AuthenticationError("Token expired")

Fix excessive timeouts:

# Before
RESET_TOKEN_EXPIRY = timedelta(days=7)  # Too long!

# After
RESET_TOKEN_EXPIRY = timedelta(hours=1)  # Appropriate

Add rate limiting:

# Before
@app.route('/login', methods=['POST'])
def login():
    pass

# After
@app.route('/login', methods=['POST'])
@limiter.limit("5 per 15 minutes")
def login():
    pass

Fix hardcoded timeouts:

# Before
expiry = datetime.utcnow() + timedelta(seconds=3600)  # Magic number

# After
SESSION_TIMEOUT = timedelta(hours=1)  # Named constant
expiry = datetime.utcnow() + SESSION_TIMEOUT

Quick Reference

Critical Vulnerabilities

Missing expiration check:

• Token/session has expiry field but it's never validated
• High risk: Expired tokens still work

No rate limiting:

• Authentication endpoints without throttling
• High risk: Brute force attacks

Client-side only validation:

• Expiration checked in frontend but not backend
• Critical risk: Easily bypassed

Excessive timeouts:

• Reset tokens valid for days
• Sessions never expire
• Medium risk: Extended attack window

Detection Patterns

Python:

# Look for
jwt.decode(..., verify=False)  # Bad
timedelta(days=7)  # Check context
@limiter.limit  # Good
datetime.now() vs datetime.utcnow()  # Timezone issue

JavaScript:

// Look for
jwt.decode(...)  // Check if expiration validated
Date.now() < exp  // Client-side check
rateLimit(...)  // Good
setTimeout(..., 86400000)  // Magic number

Java:

// Look for
Jwts.parser()  // Check if expiration validated
Duration.ofDays(30)  // Check context
@RateLimit  // Good

Helper Script

The check_intervals.py script automates detection:

# Check Python code
python scripts/check_intervals.py src/ --language python

# Check JavaScript code
python scripts/check_intervals.py src/ --language javascript

# Auto-detect language
python scripts/check_intervals.py src/

Output provides:

• Categorized issues by severity
• File and line number
• Issue description
• Fix recommendation

Best Practices

Always:

• Set expiration times for all security tokens
• Validate expiration server-side
• Use reasonable timeout values
• Implement rate limiting on auth endpoints
• Use UTC for all timestamps
• Use named constants for timeouts
• Document timeout decisions

Never:

• Skip expiration validation
• Use client-side only checks
• Set excessive timeout values
• Use magic number timeouts
• Mix timezone-aware and naive datetimes
• Use non-constant-time comparisons for secrets