Game Localizer handles local files, model-provider credentials, publishing credentials, translation memory, and release artifacts. Security reports are therefore treated as part of release quality, not as ordinary feature requests.

Security fixes are applied to the latest development branch and, after tagged releases begin, to supported release lines when the issue materially affects users of those versions.

Please do not open a public issue for a vulnerability that could expose credentials, allow unintended file access, bypass release authorization, enable unsafe command execution, or compromise published artifacts.

Instead, use GitHub's private vulnerability reporting / security advisory flow for this repository when available. If that channel is unavailable, contact the maintainer privately through the contact method on the maintainer's GitHub profile.

Please include, when possible:

• affected version or commit;
• impact and attack prerequisites;
• minimal reproduction steps;
• affected operating system and workflow;
• whether credentials, local paths, translation memory, build outputs, or remote publishing are involved;
• any suggested mitigation.

Do not include real credentials, private localization data, or copyrighted game assets in a report.

The project follows these principles:

• provider and publishing secrets must not be committed to the repository or written into project prompts, translation memory, or routine logs;
• local services should bind to loopback by default unless an operator explicitly configures otherwise;
• paths supplied through UI, configuration, or future desktop integrations must be validated before file operations;
• model-assisted decisions do not replace deterministic checks for placeholders, terminology, formatting, QualityGate, or release governance;
• agentic workflows must use allowlisted application tools rather than arbitrary shell, unrestricted path access, raw SQLite access, or arbitrary code execution;
• applying translation-memory changes, creating release artifacts, and publishing to remote targets must preserve explicit validation and authorization boundaries;
• security-sensitive failures should fail closed rather than silently downgrade protections.

CI performs full-history secret scanning. If a credential is committed, removing it from the current tree is not sufficient: revoke or rotate the credential first, then treat repository-history cleanup as a separate remediation step.

The maintainer will aim to acknowledge valid reports, assess severity and affected versions, prepare a fix or mitigation, and coordinate disclosure proportionally to the issue. Exact response times are not guaranteed for this volunteer-maintained project.