Privacy and data handling

Manuscripts are sensitive creative material. Treat them as private source documents.

General guidance

Only upload or import material that you are comfortable processing in the current environment. Do not include:

• passwords;
• API keys;
• private keys;
• legal secrets;
• unrelated personal data;
• confidential third-party material you do not have permission to process.

Local and provider processing

Depending on how Arc is configured, source text may be processed locally or sent to a configured AI provider for semantic analysis. Do not assume every workflow is fully local unless the specific environment says so.

If an external provider is used, that provider’s terms and privacy policy matter.

Evidence and derived data

Arc may produce derived project data such as:

• extracted entities;
• relationships;
• source maps;
• evidence excerpts;
• review items;
• graph files;
• story-bible material.

Treat derived data as sensitive too. Even if it is not the full manuscript, it may reveal plot, character, or world details.

Good practice

• Use small test samples first.
• Keep backups of original manuscripts.
• Do not import secrets.
• Review what files are staged before ingestion.
• Remove test data you no longer need.
• Do not share generated project folders publicly unless they contain only material you intend to disclose.

Public release note

Public hosting, account data handling, retention, deletion, and provider configuration should be documented explicitly before broad release. Until then, treat preview documentation as guidance, not a formal privacy policy.