Security is far cheaper to build in than to add after an incident. Most breaches do not involve clever, unknown attacks; they exploit well-known mistakes that were never fixed. The practices below cover the ground that matters most.
Never trust input
Everything that reaches your server — form fields, URLs, headers, uploaded files — can be manipulated.
- Validate input on the server, even if the browser already checked it
- Use parameterized queries so user input can never become SQL
- Escape output to prevent cross-site scripting
- Restrict file uploads by type and size, and store them outside the web root
Get authentication right
- Hash passwords with a modern algorithm such as Argon2 or bcrypt
- Offer multi-factor authentication, and require it for admin accounts
- Rate-limit login and password-reset attempts
- Use secure, HTTP-only cookies for sessions and expire them sensibly
Check authorization on every request
Logging in proves who someone is. It does not prove what they may do. Verify permissions on the server for every action, and never rely on a hidden button or an unguessable URL.
Protect data
- Serve everything over HTTPS
- Encrypt sensitive data at rest
- Collect only what you need — data you never stored cannot leak
- Keep secrets out of your code repository
Keep dependencies healthy
Your application includes every package it depends on.
- Update dependencies regularly
- Run automated vulnerability scans in your build pipeline
- Remove packages you no longer use
Add protective headers
A few response headers block whole categories of attack: a Content Security Policy, strict transport security, and settings that prevent your pages being framed by other sites.
Log, monitor and back up
- Record logins, permission changes and failed access attempts
- Alert on unusual patterns
- Take regular backups and test restoring them — an untested backup is only a hope
Plan for the bad day
Decide in advance who responds to an incident, how you will communicate with customers and how you would restore service. A short written plan turns panic into a checklist.
Security is a habit
No application is ever finished being secured. Review code with security in mind, test regularly and treat every fix as a chance to prevent the same class of problem elsewhere.
Written by
David Kim
Project Manager
David is a security specialist who audits web applications and helps companies implement robust security practices.
