Rotli is a local-first desktop application. Privacy and user-file integrity are product boundaries, not optional hardening.

Do not open a public issue for a suspected vulnerability or include private notes, credentials, absolute memex paths, or exploit details in a public thread.

Report it privately through GitHub private vulnerability reporting: the repository's Security → Report a vulnerability form (https://github.com/SethMed7/rotli/security/advisories/new). Only the owner sees the report.

Include only what is necessary:

• affected Rotli version and macOS version;
• the security property or boundary involved;
• minimal synthetic reproduction steps;
• expected and observed result;
• impact and required user interaction; and
• whether the issue is already public or actively exploited.

Use synthetic files. Never attach a real memex, note, chat, board, document, Keychain value, provider token, updater key, or production credential.

Security-sensitive areas include:

• secure-note and secret-shaped-content exclusion;
• filesystem roots, traversal, symlinks, permissions, and write lanes;
• provider endpoint locality, prompt/tool egress, SSRF, and CLI sandboxing;
• workspace CLI/MCP authorization, revision protection, and content filtering;
• Keychain access and credential handling;
• Tauri IPC, capabilities, CSP, webview-to-Rust validation;
• updater signatures, Developer ID signing, notarization, and release provenance; and
• managed Breve scheduling and external delivery boundaries.

The full model is docs/security/threat-model.md. The executable egress inventory and current accepted findings are in docs/development/security.md.

Security fixes target the current main branch and the newest signed release. Older releases do not receive backports; update to the newest signed build after reviewing its release notes.

No response-time or remediation SLA is promised; Rotli is maintained by one person. A confirmed issue is prioritized by user-data exposure, integrity impact, reachability, and exploitability.

Rotli coordinates disclosure after a fix and safe update path exist. A security fix includes a failing regression or adversarial fixture, the appropriate independent Rust/TypeScript guard, updated threat/egress documentation, and a review of adjacent call sites. Release evidence follows docs/operations/release-and-supply-chain.md.