SemantiqWall
PT EN
Sign in I'm a vendor

← Trust Center

Secure Development and Change Management

Version
1.0
Owner
Founder
Approved by
Murilo Martins (Founder) · 09/28/2026
Next review
2027-09-28 (annual or on significant change)
Scope
The SemantiqWall application code, SDKs, database schema, server configuration and the assets (servers, domains, supplier accounts) that make up the production service.

1. Secure development lifecycle

Phase What happens
Requirements Each change states what it does and its security impact: new data collected, new supplier, new public endpoint, change to authentication, tenancy or audit. Changes with security impact are noted in the security register.
Design The secure-coding baseline (section 2) applies. New data collection is checked against policy 08 (minimization, classification, retention).
Build Code lives in a private GitHub repository. Secrets are never committed.
Test The full automated test suite passes before every push (section 3).
Review Every change is reviewed as a diff before it is committed (section 4).
Deploy Only through the deploy script (section 6).
Operate Audit log integrity is checked every 10 minutes and at the end of each deploy; problems follow policy 05.

2. Secure-coding baseline (in place)

These rules are written in the repository's developer guide and enforced by tests:

3. Testing and acceptance criteria

4. Code review

5. Dependencies

6. Deployment

7. Rollback

  1. Identify the last good commit (git log on the server).
  2. If the bad release included migrations, roll them back first with php artisan migrate:rollback --step=N, only when the down() steps do not lose data; otherwise restore from backup (policy 06).
  3. Check out the last good commit on the server (git checkout <commit>), then run the remaining deploy steps (composer install, caches, worker restart, PHP-FPM reload, audit-chain check).
  4. Fix forward on main, then return the server to main with a normal deploy.
  5. Record the rollback in the register. The procedure is exercised at least once a year (first test: owner Founder, target 2026-12-31).

8. Server configuration changes and baseline

The server baseline is: Ubuntu 24.04 LTS with unattended security upgrades; ufw allowing only 22 (SSH), 80 (HTTP, redirected to HTTPS) and 443 (HTTPS); SSH by key only with password authentication disabled; fail2ban on SSH; NTP synchronized; Caddy, PHP 8.4-FPM, PostgreSQL 16 (SSL on) and Redis, whose ports are not open in the firewall. Commitment: confirm that PostgreSQL and Redis listen on the local host only and record it (owner: Founder, target 2026-10-15).

9. Emergency changes

When a fix cannot wait (active exploitation, outage): the change may skip the normal wait but never skips the automated tests, unless the service is down and the change is a rollback. It is recorded as an exception (policy 00) within 2 business days, with a follow-up review.

10. Changes affecting customers

Changes to customer tenants (policies, keys, settings) are made only by the customer, or by the operator at the customer's written request, and are audited. Platform changes that affect how customers use the service are announced before release; security fixes may ship immediately.

11. Asset changes

The security register holds the asset list: servers, domains, databases, supplier accounts, keys (policy 02), endpoints (policy 10) and the code repository, each with its owner, the highest data class it holds (policy 08) and its location. The list is reviewed at least once a year.