security-checklist

Files (1)

• SKILL.md

SKILL.md (first lines)

# Security Checklist

Shared security policy. `$goldband review code` with an explicit security scope
owns deep security review.

## Scope

Use this skill when changes touch:

- Authentication, authorization, tenancy, roles, sessions, or service accounts.
- User-controlled input, uploads, parsing, deserialization, templates, SQL, shell,
  or external fetches.
- Secrets, tokens, PII, payment data, medical data, audit logs, or telemetry.
- Deployment exposure, CORS, CSP, TLS, CI secrets, or destructive admin actions.

## Workflow Handoff

Independent review of code changes uses `$goldband review code` with an
explicit security scope when installed. Threat modeling without a code change,
local defensive checks, and code-change review without the workflow apply the
baseline below within the requested scope.

## Baseline Review

1. Identify the trust boundary, actors, sensitive assets, and permitted actions.
2. Confirm authentication is present on every protected route and background
   action.
3. Confirm authorization checks tenant, resource owner, role, and action. Deny
   by default when context is missing.
4. Validate external input at the boundary with schemas or parsers.
5. Keep secrets out of source, logs, telemetry, screenshots, and fixtures.
6. Use parameterized APIs for SQL, shell, templates, and external fetches.
7. Make security failures explicit, auditable, and testable.

## Completion Check

Before considering security work complete, state:

- The trust boundary reviewed.
- The controls verified.
- The tests, scanners, or inspections run.
- Any security assumptions that remain unverified.

From leo110047/goldband, MIT.

More from leo110047/goldband