Skip to content

Security: merlinguild/transparency

Security

SECURITY.md

Security policy

This policy covers Merlin Guild and its distribution infrastructure: the bootstrap installer, the signed release manifest, the download service, and the advisory feed.

Reporting a vulnerability

We welcome reports from security researchers, integrators, and users.

Contact

  • Email: security@merlinguild.com. PGP-encrypted reports are preferred.
  • PGP key: merlinguild-security-pgp-2026-05.pub.asc (Ed25519 / Curve25519) in the repository root.
  • security.txt: per RFC 9116, available at merlinguild.com/.well-known/security.txt.
  • Fallback: if the primary contact is unavailable for more than 7 days, use the contact listed in security.txt.
  • Please do not open public issues or discuss report details publicly before the coordinated disclosure window closes.

What to include

A useful report contains:

  • Affected version (CalVer, e.g. 26.4.1).
  • Affected component (cryptographic core, distribution pipeline, license verification, advisory feed, etc.).
  • Reproduction steps or proof-of-concept.
  • Impact assessment.
  • Suggested remediation, if any.
  • Whether you intend to publish; preferred coordinated disclosure timeline.

Response targets

These are targets for a solo maintainer, not contractual SLAs. There is no paid bug bounty at this time.

Stage Target
Acknowledge receipt. 72 hours.
Triage and severity assessment. 7 days.
Fix for CRITICAL or HIGH. 7 days from triage, best effort.
Fix for MEDIUM. Next scheduled release.
Fix for LOW. Best effort, next feature release.
Coordinated disclosure window. 90 days from initial report.

If a vulnerability is actively exploited in the wild, the timeline collapses: fix and notification within 24 to 72 hours.

Safe harbour

We will not pursue legal action against researchers acting in good faith who:

  • Do not access user data beyond what is necessary to demonstrate the vulnerability.
  • Do not modify or destroy data.
  • Do not disrupt service for other users.
  • Disclose responsibly per the timeline above.

Supported versions

Merlin Guild ships an evergreen rolling release: a single current build, no maintenance branches, no backports.

Status Definition Support
Current Latest release. Full security and feature support.
Older Any earlier release. No dedicated fixes; install the latest release, which stays free for security updates.

Vulnerability handling process

  1. Receipt and acknowledgement: within 72 hours.
  2. Triage: within 7 days; severity per CVSS 3.1 and exploit status.
  3. Fix development: root cause analysis, regression test, and a VEX statement for affected dependency CVEs where applicable.
  4. CVE assignment: for HIGH and CRITICAL, via GitHub Security Advisory.
  5. Release and signing: the fix ships in the next release; all artifacts are signed.
  6. User notification: an entry is added to the signed advisories.json feed, published simultaneously to the download host and to this repository. Critical advisories surface as a blocking dialog in-app.
  7. Public disclosure: the full advisory is published in advisories/MG-SEC-YYYY-NNNN.md after the coordinated disclosure window, or earlier if mutually agreed.
  8. Reporter acknowledgement: credit in the advisory, with consent.

Verifying releases

The installer verifies before it executes: install.ps1 checks the detached signature of manifest.json against the pinned manifest public key, then the SHA-256 of the downloaded installer against the manifest. Release public keys (license, advisory, manifest) live in the repository root. The advisory feed can be verified with verify-advisory.ps1.

Out of scope

  • Defects in the operating system or hardware that bypass application-level cryptography.
  • Active malware, keyloggers, or screen capture on the device while the vault is unlocked.
  • Theft of the master password through social engineering.
  • Loss of the recovery key combined with loss of the original device.
  • Findings in third-party dependencies declared not_affected in our VEX document.
  • Findings in test infrastructure, the marketing site, or internal tooling not shipped to users.
  • Attacks that already require arbitrary code execution on the device.

These reports are still welcome: they inform hardening guidance, but they do not trigger the response targets above.

There aren't any published security advisories