Skip to content

Security: heidihelena/notevahti

Security

SECURITY.md

Security policy

NoteVahti processes data abstracted from clinical notes, so security and data protection are first-order concerns. Please read this before reporting.

Never include PHI or real patient data

Do not put real patient data, PHI, or identifiable note text in issues, pull requests, discussions, or vulnerability reports. Reproduce problems with the synthetic corpus (corpus/) or with fabricated examples. Reports containing real patient data will be deleted.

Security posture (what NoteVahti does and does not do)

  • The validation core is local-first and offline: it makes no network calls (there is a test asserting this), sends no telemetry, and has no runtime dependencies. PHI does not leave the process.
  • Note text and matched snippets are hashed by default in the audit record; retaining raw text is an explicit local opt-in.
  • The audit log is append-only and hash-chained; tampering with a past entry is detectable (AuditLog.verify). Note: this is tamper-evident, not tamper-proof, and the hash chain proves integrity, not the identity of who wrote an entry.
  • Out of scope: any extractor, model, or pipeline you plug in behind the Extractor interface. Those may have their own network/PHI behaviour — that is your responsibility, not NoteVahti's.

Reporting a vulnerability

Please report privately, not as a public issue:

  • Preferred: GitHub "Report a vulnerability" (the Security tab of this repository → Private vulnerability reporting).
  • Include: affected version/commit, a synthetic or fabricated reproduction, impact, and any suggested fix. No real patient data.

We aim to acknowledge a report within a reasonable time and will coordinate a fix and disclosure. There is no bug-bounty programme.

Supported versions

Pre-alpha (0.1.x.devN). The API and behaviour may change without notice and there are no security or support guarantees yet. Use at your own risk; this is not a medical device and not clinical advice.

There aren't any published security advisories