What is the first pass costing your firm? Run your numbers
Trust centre Trust

Secure development

Version 1.1 · October 2026

1. Purpose

This policy describes how Solicia develops, tests, and ships software securely. It is informed by the OWASP Software Assurance Maturity Model (SAMM) and scaled to the size of the company. The controls below are automated and enforced by tooling, so they cannot be skipped under time pressure.

2. Change management

  • All changes reach production via pull request. Direct pushes to the main branch are not used. Every change is recorded, attributable and reviewable.
  • Continuous integration gates every pull request:
    • Full automated test suite (16,000+ tests) must pass
    • Golden-regression suite guards core output quality against behavioural drift
    • Secrets scanning (gitleaks) blocks credentials from entering source control
    • Static application security testing (Semgrep) blocks on error-level findings
    • Container image vulnerability scan (Trivy) blocks on fixable critical and high findings
    • Pull-request hygiene checks (structured description validation)
  • Branch protection on the main branch requires these checks and a review. Merges are squash-only and attributable.
  • Changes are reviewed before merge, including AI-assisted code review for defects and security issues.

3. Dependency and vulnerability management

  • Dependencies are audited against known-vulnerability databases on every build (`pnpm audit`). The build fails on any high or critical advisory without a documented, time-limited acceptance.
  • GitHub Dependabot alerts and security updates are enabled, with weekly grouped dependency update pull requests for npm packages, GitHub Actions and the container base image.
  • The production container image is scanned on every pull request and weekly. Development dependencies and build tooling are pruned from the runtime image, which runs as a non-root user.
  • Remediation targets: critical within 7 days, high within 30 days, moderate within 90 days or documented acceptance with rationale in the quarterly Internal Security Review.
  • A scheduled dependency upgrade pass runs at least quarterly.

4. Environments and data

  • Development, test, and production are separate environments with separate credentials.
  • Customer data is never used in development, test, or QA. Non-production environments run synthetic data only.
  • Production secrets exist only in the production platform's secret store; they are never present in source control, CI logs, or developer machines beyond operational necessity.

5. Build and deployment

  • Production artifacts are immutable, versioned container images built in a controlled registry (Azure Container Registry).
  • Deployments are revisioned (Azure Container Apps); any release can be rolled back to the previous known-good revision.
  • Infrastructure is defined in code (Terraform) and reconciled against live state.

6. Secure design practices

  • Tenant data is isolated by tenant-scoped storage prefixes with programmatic enforcement (cross-tenant path access raises an error by construction).
  • Application credentials are stored as salted adaptive hashes.
  • Inbound webhooks are signature-verified; outbound origins are locked to the edge provider (requests cannot bypass the TLS edge).
  • For new features, data residency and tenant isolation are treated as design constraints.

7. Review

This policy and its controls are assessed in the quarterly Internal Security Review; customers may review this policy and the associated review reports on request.