goraz v0.2.0

Comparison of existing secret-management projects

Research date: 2026-11-06. This document records what goraz learned from earlier projects. It is a summary of published material, not an assessment of those projects' overall quality.

How to read the evidence

  • A linked claim comes from a page that was read during this research.
  • (secondary) means the claim comes from a third-party article or a search excerpt, not from the project or the original researcher.
  • (unconfirmed) means no page that was read supports the claim. Do not cite it without checking.
  • "None found" means only that the searches found nothing. It does not show that nothing exists.
  • CVE scores and fixed versions vary between sources. Check the original advisory before relying on one.

1. The findings that shaped the design

#FindingEvidenceConsequence for goraz
1Storage and servers must be treated as hostile. Researchers presented 12 attacks on Bitwarden, 7 on LastPass, 6 on Dashlane and 6 on 1Password under a malicious server. Causes include unauthenticated metadata and KDF settings, missing key separation, and legacy code paths.ETH Zurich and USI, ePrint 2026/058, USENIX Security 2026 versionAllowlisted KDF parameters, every ciphertext bound to its context, all metadata encrypted, no legacy path
2Most recent incidents are not cryptographic failures. They are parser, path, canonicalization and authorization bugs.Vault CVE-2026-3605, CVE-2025-6014; gopass GHSA-v4c2-4fvx-chr3; SOPS GHSA-jgf3-f6rg-8x3hOne canonicalizer, strict parsers, fuzzing, and a single-file store with no path derived from data
3Administrative features become code execution.Vault CVE-2025-6000 (secondary) and its bypass CVE-2026-5051; Conjur CVE-2025-49828; OpenBao GHSA-j6wc-jpvg-xfxqNo plugins, no template evaluation, no log path or binary path taken from data in the first release
4Audit redaction regresses.Vault CVE-2024-8365; OpenBao GHSA-ghfh-fmx4-26h8Typed audit events with no free-form field, plus canary tests in CI
5Revocation and sealing are often weaker than they sound.Almond audit found Vault keys in memory after sealing; Infisical may accept a revoked token for up to 12 minutes during a Redis outage; OpenBao AppRole secret IDs outlived expiry until tidy ranCheck expiry and revocation at each use, fail closed, and state the limits plainly
6Secrets tools are prime supply-chain targets.Bitwarden CLI 2026.4.0 on npm, KeeLoader, LastPass 2022Few dependencies, static binaries, no install scripts, signed and attested releases
7Honest documentation of non-goals is a security feature.1Password's "Beware of the leopard", gopass security.md, Vault security modelA published list of what is not defended, and no use of the phrase "zero knowledge"

2. Summary table

ProjectKindLicenseData encryptionKDF or root protectionMetadata protectedPublic third-party audit
HashiCorp VaultServerBSL 1.1AES-256-GCMShamir or KMSStored data, yesAlmond for ANSSI, 2026
OpenBaoServerMPL-2.0AES-256-GCMShamir, KMS, static keyStored data, yesNone found
InfisicalServerMIT, with ee/ separately licensedAES-256-GCMOperator key or HSMStored data, yesPen tests stated, reports on request
DopplerSaaSCommercial; details unconfirmedAES-GCM, per-workplace keyHSM-backed KMSSelf-reportedNone found
CyberArk Conjur OSSServerLGPL-3.0 serverAES-256-GCM (Slosilo)CONJUR_DATA_KEY environment variableNot establishedUnsupported README claim
SOPSFile toolMPL-2.0AES-256-GCM per valueAge, PGP or cloud KMS wrapNo, keys are cleartext by designNone found
ageFile format and toolBSD-3-ClauseChaCha20-Poly1305 STREAMX25519, ML-KEM-768 hybrid, scryptNot applicableNone found
passCLIGPL-2.0-or-laterGPGGPG keyNo, names are file namesNone found
gopassCLIMITGPG by default, age experimentalGPG or age identityNoNone found
TellerBroker CLIApache-2.0Delegated to backendDelegatedDelegatedNone found
BitwardenPassword managerAGPL-3.0 server, with a separately licensed partAES-256-CBC with HMAC-SHA-256PBKDF2 600,000 or Argon2id 32 MiB, t=6, p=4PartlySeveral, listed by vendor
VaultwardenServer for Bitwarden clientsAGPL-3.0Client sideClient sideAs BitwardenNone found
KeePassXCPassword managerGPL-3.0AES-256-CBC or ChaCha20 with HMAC blocksArgon2, 64 MiB, 10 rounds, p=4 constantsYes, whole databaseMolotnikov, 2023
1PasswordPassword manager and CLIUnconfirmed; white paper publicAES-256-GCMPBKDF2 650,000 plus 128-bit Secret KeyItem overview and details, yesClaims not reviewed here
go-keyringOS keyring libraryMITDelegated to OSOSOSNone found
99designs/keyringOS keyring libraryMITDelegated to OS or JWT fileOS or passphraseOSNone found

3. Server and platform managers

HashiCorp Vault

OpenBao

Infisical

Doppler

CyberArk Conjur

4. File and CLI tools

SOPS

  • Worth keeping: value-level encryption with the key path as AEAD additional data, key groups with Shamir, and the separation of updatekeys from rotate.
  • Threat model: the security page orders risks by likelihood, from compromised cloud credentials to weak AES. It does not discuss integrity.
  • Integrity detail, confirmed in sops.go: the MAC is SHA-512 over the walked leaf values, written with no length framing. Key names and recipient metadata are not written to that hash within Tree.Encrypt.
  • Incidents:
    • GHSA-jgf3-f6rg-8x3h (High, 7.4): decrypting an untrusted file sends the user's Vault token to the address named in that file. Lesson: file metadata must never select a network endpoint that receives a credential.
    • GHSA-x4cm-pcq4-39j4: exec-file --filename ../x writes plaintext outside the temporary directory.
    • Issue #2243: a refactor changed MAC input and broke existing files. Lesson: authenticate exact bytes, and freeze formats with golden vectors.

age

pass

gopass

Teller

Others reviewed briefly

  • git-crypt: no revocation; an edit to .gitattributes can turn off encryption. Lesson: fail closed.
  • transcrypt: unauthenticated CBC. An example to avoid.
  • dotenvx: good developer experience; look-alike packages exist (secondary).
  • chamber: explicit precedence rules for exec.
  • Sealed Secrets: the clearest explanation of key renewal versus secret rotation.
  • Passbolt: published Cure53 results, including a High finding that clients do not verify metadata signatures. Lesson: signing without verifying gives no integrity.
  • Akeyless: novel fragment cryptography. Lesson: new constructions need independent review.

5. Password managers

Bitwarden

Vaultwarden

  • Worth keeping: a small self-hosted server.
  • Threat model: vault cryptography lives in the Bitwarden clients.
  • Record: the October 2026 advisories include a revoked member keeping organisation access, and 2FA remember tokens that survive an administrator removing 2FA.
  • Lesson: end-to-end encryption does not make server authorization irrelevant.

KeePassXC and KDBX 4

1Password

LastPass (as a lesson)

6. Go building blocks

OS keyring libraries

zalando/go-keyring99designs/keyringByteNess/keyring
Latest versionv0.2.8 (2026-03)v1.2.2 (2022-12)v1.15.1 (2026-11)
StatusMaintainedMaintenance question unansweredMaintained fork
macOS path/usr/bin/security, no cgoSecurity framework, cgoSecurity framework, cgo

Go memory handling

  • The Go garbage collector is non-moving, but goroutine stacks are copied on growth, and libraries copy key material.
  • runtime/secret requires GOEXPERIMENT=runtimesecret, carries no compatibility promise, and does nothing outside Linux amd64 and arm64 in the released documentation.
  • x/crypto/argon2 allocates its matrix on the ordinary heap, has no wiping code and panics on invalid parameters. RFC 9106 advises memory wiping where side channels are a concern, so this is a known gap.
  • memguard offers off-heap locked memory, but key bytes still enter the Go heap when passed to standard cipher packages.

Parameters and standards

SourceRecommendation
RFC 9106 firstArgon2id, t=1, p=4, 2 GiB, 128-bit salt
RFC 9106 secondArgon2id, t=3, p=4, 64 MiB
OWASPMinimums for login verifiers, such as 19 MiB, t=2, p=1. Not suitable for an offline vault file
NIST SP 800-63B-4Minimum 15 characters for single-factor passwords, no composition rules, blocklist, NFC normalization, paste allowed
Len et al., 2021Non-committing AEADs enable partitioning oracles
Albertini et al., 2022Generic key commitment through two derived keys
GitHub token formatIdentifiable prefix and CRC32 checksum
GitHub, 2014SHA-256 digests for high-entropy tokens at rest, because a slow KDF is unnecessary for them

Current versions checked on 2026-11-06

Go 1.27.3 and 1.26.10, golang.org/x/crypto v0.58.0, filippo.io/age v1.3.2, golang.org/x/vuln v1.9.0, gosec v2.30.0, staticcheck 2026.2.2.

7. Gaps in this research

  • NVD pages rendered empty, so several CVE details come from project advisories or aggregators.
  • Cyata's and CyberArk's own write-ups on Vault Fault could not be opened, so that chain rests on secondary reporting.
  • OpenBao advisory list pages 4 and 6 were not read.
  • The Cure53 reports on Bitwarden were not read in full.
  • No formal threat model was available for Doppler, Conjur or Teller, and no audit report was available for many others. A "None found" entry should not be read as a criticism.

Edit this page on GitHub