• Define who owns security decisions and policy maintenance.
• State which environments, services, and contributors this policy covers.

• Store secrets only in approved secret stores or local .env files (never commit credentials).
• Use least-privilege credentials and separate keys by environment.
• Rotate credentials on schedule and immediately after suspected exposure.
• Do not place tokens, API keys, or private credentials in logs, prompts, screenshots, or tickets.

• Define data classes (public, internal, sensitive, restricted) and handling rules for each.
• Grant data access by role and minimum required scope.
• Restrict direct production data access; prefer audited access paths.
• Define retention/deletion expectations for logs, prompts, and model outputs.

• Define disallowed AI use cases and prohibited output categories.
• Define allowed model/provider list and escalation path for exceptions.
• Require validation/guardrails for model outputs before high-impact actions.
• Define when human review is mandatory (for example legal, financial, safety-critical outputs).

• Separate dev/stage/prod credentials, data, and access controls.
• Require auditable change history for security-sensitive configuration.
• Document baseline controls for network access, dependency updates, and runtime hardening.

• Define vulnerability reporting channel and response ownership.
• Define triage priorities and containment expectations.
• Document communication and remediation follow-up requirements.

List applicable regulatory, contractual, or internal compliance requirements.