Skip to content

Security: commercekit-tools/action-integration-tests

Security

SECURITY.md

Security policy

This public repository contains synthetic integration fixtures and test code. Real SAP credentials, Cloudflare secrets, webhook URLs, access tokens, and customer content must never enter committed source, logs, summaries, artifacts, or issues.

Sandbox run creation returns temporary control and OAuth credentials. The harness masks them immediately, retains them only in private runner-temporary files and same-job environment values, and attempts to delete the run in always-run cleanup. Hard runner termination can prevent cleanup; the Sandbox's 24-hour expiry is the fallback. Expiry does not replace cleanup assertions.

Slack and Teams webhook URLs are credentials. Configure only dedicated test destinations in the provider-smoke environment. Live notifications require a manual dispatch and their own confirmation. Check environment protection before dispatch; creating an environment in YAML does not configure required reviewers or branch restrictions.

GitHub Packages uses the job's automatic read-only GITHUB_TOKEN for dependency installation. Do not add a long-lived package PAT as a repository secret. Local contributors keep read credentials in user-level or ephemeral authentication configuration outside this repository.

The control API and Management API have different authorization schemes. Keep control tokens scoped to their run, use the existing shared client for Management requests, and preserve fixed-origin checks, bounded reads, timeouts, and redirect rejection. Never weaken provider URL validation to send Slack traffic to a generic mock endpoint.

Report vulnerabilities through GitHub private vulnerability reporting when available, or GitHub's private security advisory reporting flow. Do not post credentials or exploit details in public issues. If a test webhook or generated credential is exposed, revoke/delete the destination or run and follow the organization's incident procedure.

There aren't any published security advisories