Secure SDLC

• Security review for security-sensitive changes.
• Threat modeling for auth, integrations, AI, evidence, and testing workflows.
• Code review for production changes.
• Dependency scanning and dependency review.
• Secrets scanning.
• Schema validation at system boundaries.
• Testing where practical, including security-focused regression coverage.
• Release checks before deployment.
• Human review of AI-assisted code and content.
• Vulnerability intake, triage, and remediation.
• Documentation updates for trust, legal, and security-impacting changes.

These changes receive elevated review because they can affect authorization, evidence handling, payments, or customer trust:

• Auth and session changes.
• Organization, workspace, role, and entitlement changes.
• Legal, finance, IT, and security delegation changes.
• SSO, SAML, OIDC, and SCIM changes.
• Stripe, checkout, and seat provisioning changes.
• Private-offer, SOW, and contract workflow changes.
• Packet, report, and evidence rendering changes.
• Workbench Copilot changes.
• Model-provider, prompt, retrieval, or workspace-context changes.
• Browser extension changes.
• Native app changes.
• Connector and OAuth scope changes.
• Runtime tracing, model gateway, or trace-handling changes.
• Code scanner, adversarial range, or RAG harness changes.
• Database migrations and RLS or policy changes.
• Uploaded-file and artifact-handling changes.
• Trust-center and legal route changes.

We review design changes for the failure modes that matter most to this product line:

• Tenant and workspace isolation.
• IDOR and broken authorization.
• Role and entitlement escalation.
• Evidence and artifact exposure.
• Private-offer and contract packet exposure.
• Prompt and evidence leakage.
• Connector and OAuth scope abuse.
• Copilot misuse or overreach.
• Payment and seat-provisioning abuse.
• Testing authorization and ROE boundary failures.
• Public claim and attestation misuse.
• Unsafe handling of target details, logs, traces, prompts, or reports.

• TypeScript and schema validation at the boundary where data enters the system.
• Explicit authorization checks on the server side.
• Privileged secrets handled server-side only.
• No service-role keys in browser code.
• Least privilege and secure defaults.
• Safe error handling and logging without secrets.
• Input validation and output encoding where relevant.
• Dependency review before adoption and release.
• Secure handling of migrations.
• Review of generated or AI-assisted code before merge.

• AI may assist with drafting, code review, analysis, refactoring, tests, and documentation.
• AI-generated code and content require human review.
• Security-sensitive paths require extra scrutiny.
• Customer secrets and restricted customer evidence should not be pasted into AI coding tools unless the applicable agreement and processing path permit it.
• AI assistance does not replace security review.

• Unit tests where useful.
• Integration tests for API and state transitions.
• Browser and end-to-end tests for customer journeys.
• Auth and role tests.
• /scope and /start funnel state tests.
• Anonymous-to-authenticated continuation tests.
• Stripe test mode for checkout and entitlements.
• Packet generation tests.
• Contract, ROE, and evidence flow tests.
• SSO and SCIM fixtures where available.
• Regression tests for security-sensitive routes.

• Preview or staging review where available.
• Feature flags if used.
• Migration review before deployment.
• Rollback planning for security-sensitive releases.
• Monitoring after sensitive releases.
• Route and metadata updates when product surfaces change.
• Updating Trust Center docs when product behavior changes.
• Reviewing customer-facing legal and security copy when capabilities change.

• Dependency updates.
• Vulnerability disclosure intake.
• Triage and severity assessment.
• Remediation tracking.
• Verification and retest.
• Customer notice where appropriate.
• Lessons learned flow back into the SDLC.

The following changes require elevated review before release:

• New AI provider or model route.
• New OAuth scope or connector.
• New browser extension permission.
• New native app capability.
• New way to ingest or upload evidence.
• New way to generate public-facing claims.
• New payment or entitlement workflow.
• New admin or role permission.
• New red-team or pentest workflow.
• New cloud or testing boundary feature.