DueFlow is local-first and handles user-provided notices, files, schedules and optional model-provider credentials. Please do not file public issues containing private user data, API keys, local database files or sensitive diagnostics.

DueFlow is currently pre-1.0. Security fixes target the main development branch until stable release branches exist.

If you find a vulnerability, please report it privately to the project maintainers. If a private security advisory channel is available on GitHub, use that first. Otherwise, contact the maintainers listed in the repository metadata.

Please include:

• affected version or commit;
• operating system;
• reproduction steps;
• expected and actual behavior;
• whether the issue can expose local files, API keys, Inbox content, task data, backups or diagnostics;
• any suggested fix or mitigation.

Do not include real API keys, private Inbox text, full local databases, or personal schedules in the report. Use synthetic sample data whenever possible.

Current intended boundaries:

• Model output is not committed as tasks until the user confirms drafts.
• Desktop diagnostics omit raw Inbox text, task titles, source quotes and extracted task descriptions.
• Local pet appearances are imported from manifest-declared local assets only, reject URLs and parent traversal, and use staging for rollback.
• Local skill manifests are declarative and read-only; they are not executed as third-party code.
• Runtime secrets are read from local environment variables and .env, which must not be committed.

• Public multi-user hosting.
• Account authentication and authorization.
• Cloud sync.
• Automatic reading of private messaging apps.

These may become in scope if the project adds hosted or multi-device features.