
Security Model
The plugin's full security model: server-only validation, single-use challenges, CSRF, and rate limiting.
- Answers are validated only on the server — nothing about the challenge or the check
ever depends on client-side JavaScript being trustworthy.
- Every challenge is single-use and expires.
- Challenges are bound to the session that created them.
- The refresh (new-sum) endpoint requires a valid Joomla CSRF token.
- Wrong answers trigger a cooldown that grows with each further mistake.
- Answers and refresh requests are both capped per session.
- All input is trimmed, length-limited, and checked to be numeric-or-a-known-word before
anything else happens to it.
- All output is escaped; every administrator-supplied colour and length value is validated
against a strict pattern before being written into a page-level stylesheet.
- No
eval, no inline event handlers, no inlinestyleattributes — a strict Content
Security Policy is unaffected. The one page-level <style> block (carrying your colour and size choices) goes through Joomla's own Web Asset Manager, so Joomla's CSP plugin can add a nonce to it like anything else.
A deliberate trade-off: rate limiting counts against the visitor's session, not an IP address. That keeps the plugin free of any network identifier whatsoever, and it still stops an ordinary form-spam script (which keeps its session cookie between requests). It is not a defence against a distributed attacker who discards cookies between requests — that's a job for a firewall or reverse-proxy rule, which is the right layer for it regardless of what captcha you use.