Decision Workspace
atlas-detect vs split-brain-harness vs guardrail-core
Side-by-side comparison of Rust crates
43
atlas-detect
experimentalv0.1.0
MITRE ATLAS technique detection for LLM and AI agent security. Detects prompt injection, jailbreaks, credential exfiltration, model extraction, and 90+ other AI-specific attack techniques.
49
split-brain-harness
experimentalv1.6.0
Soul-injected two-stage LLM telemetry pipeline. Wraps any LLM with affective/intent/cognitive analysis, deterministic verification, and a Stage 0 deobfuscation normalizer. Drop-in OpenAI-compatible proxy.
50
guardrail-core
experimentalv0.1.1
Core types, pipeline, and stage trait for guardrail-rs
Core Metrics
| atlas-detect | split-brain-harness | guardrail-core | |
|---|---|---|---|
| Health Score | 43 | 49 | 50 |
| Total Downloads | 2.2K | 135 | 331 |
| 30d Downloads | 0 | 0 | 0 |
| Dependents | 0 | 0 | 8 |
| Releases | 1 | 7 | 2 |
| Last Updated | 180d ago | 6d ago | 97d ago |
| Age | 6m | 3m | 3m |
Health Breakdown
atlas-detect
Maintenance
5
Quality
15
Community
6
Popularity
4
Documentation
13
split-brain-harness
Maintenance
19
Quality
11
Community
6
Popularity
3
Documentation
10
guardrail-core
Maintenance
12
Quality
14
Community
9
Popularity
3
Documentation
12
Technical Details
| atlas-detect | split-brain-harness | guardrail-core | |
|---|---|---|---|
| Version | 0.1.0 | 1.6.0 | 0.1.1 |
| Stable (≥1.0) | ✗ No | ✓ Yes | ✗ No |
| License | Apache-2.0 | MIT | Apache-2.0 OR MIT |
| Dependencies | 4 | 19 | 11 |
| Crate Size | 20KB | 216KB | 29KB |
| Features | 2 | 4 | 2 |
| Yanked % | 0.0% | 14.3% | 0.0% |
| Edition | 2021 | 2021 | 2021 |
| MSRV | — | — | 1.75 |
| Owners | 1 | 1 | 1 |
Links
Quick Verdict
- •guardrail-core leads with a health score of 50/100, but none of the options score above 80.
- •atlas-detect has the most downloads (2.2K), suggesting wider adoption.
- •atlas-detect, guardrail-core are pre-1.0 — API may change.