Skip to content

Security: Rennlabs/grokprint

Security

SECURITY.md

Security Policy

Scope

grokprint is an unofficial local tool. It reads Grok session files under ~/.grok/sessions/ and writes orientation cards (grokprint.json, last-print.md). It does not phone home.

What we care about

  1. Secret leakage in cards — tool args and chat text can contain tokens. Digests pass through redaction (src/grokprint/redaction.py) before write. Prefer over-redact; report misses with a redacted repro fixture.
  2. Hooks fail-open — Stop hooks must never block the agent on error.
  3. No credentials in the repo — never commit .env, session dumps, or real tokens.

Reporting

Open a GitHub issue with the security label, or email the maintainers via the org contact on GitHub. Do not attach real secrets; use synthetic fixtures.

Supported versions

Best-effort on the latest main / latest release tag only. There is no paid SLA.

Threat model (short)

Asset Risk Mitigation
Session transcripts on disk Cards re-surface content Redaction; parent-only; local files only
Hook execution Malicious project hooks Grok folder-trust; this project only writes digests
Install Path confusion Prefer pip install -e . + python -m grokprint.hook

There aren't any published security advisories