Garak vs HOL Guard
HOL Guard leads on trust: 91.3/100 (Grade A) against 66.5/100 (Grade B), a 24.8-point gap. Garak leads on adoption and rests on broader evidence; HOL Guard leads on supply-chain integrity and provenance. Note: HOL Guard's evidence coverage is thinner (coverage B vs A) — its score rests on fewer independent signal types.
the LLM vulnerability scanner
Choose Garak if adoption matters most.
- +2.6Adoption: 11.5k weekly downloads against 8.7k
- A vs BEvidence coverage: 4 of 5 independent signal types, against 3
Open-source antivirus for AI agents: block risky tools, secret access, prompt injection, malicious packages, MCP servers, plugins, and skills at runtime.
Choose HOL Guard if supply-chain integrity and provenance matter most.
- +13.8Safety / Integrity: 100% of recent commits signed, against 76%
- +7.2Identity / Provenance: package provenance attested, against none
- +3.5Transparency: OSSF Scorecard 8.8 against 4.7
- +1.2Maintenance: last push today, against 2d ago
Where they differ
An independent, evidence-based trust comparison of Garak and HOL Guard, two Security & Guardrails projects in the HVTracker registry. Scores come from public, checkable signals — supply-chain provenance, OSSF Scorecard, maintenance, and adoption — not popularity.
Full evidence
| Signal | GarakNVIDIA/garak | HOL Guardhashgraph-online/hol-guard |
|---|---|---|
| HVTrust score | 66.5 | 91.3 |
| Evidence grade | B | A |
| Coverage grade | A | B |
| Overall rank | #460 | #17 |
| Rank in Security & Guardrails | #7 | #2 |
| GitHub stars | 9.6k | 926 |
| Last updated | 2d ago | today |
| Build provenance | No | Yes |
| OSSF Scorecard | 4.7 / 10 | 8.8 / 10 |
| License | Apache-2.0 | Apache-2.0 |
| Downloads | 12k/wk | 9k/wk |
| Trust dimensions (points earned) | ||
| Safety / integrity / 25 | 9.7 | 23.5 |
| Identity & provenance / 18 | 10.8 | 18.0 |
| Transparency / 17 | 12.5 | 16.0 |
| Maintenance / 20 | 18.8 | 20.0 |
| Adoption / 20 | 15.0 | 12.4 |
| Runtime capability surface (full matrix) | ||
| MCP server | — | Implemented |
| External providers | 8 — Amazon Bedrock, Anthropic, Cohere, … | 3 — Anthropic, Google Gemini, Multi-provider (LiteLLM) |
| Requires API keys | Yes | No |
| Plugin surface | plugins | extensions |
| Provenance drift | Match | Match |
How to read this: HVTrust (0–100) weighs supply-chain signals (provenance, OSSF Scorecard, signed commits, open license) alongside real-world adoption, scaled by an evidence-confidence factor. Grade bands: A ≥ 80, B ≥ 65, C ≥ 50, D < 50. Signals refresh daily. Full methodology v4.4 →