High-performance Rust backend for Hermes Agent — ONNX Runtime, Candle, Hailo-8, multi-GPU, REST API
SaferSkills independently audited hermes-rust-backend (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
]() ]() ]()
High-performance Rust backend for Hermes Agent — LocalAI proxy, MCP bridge, Hailo-8 vision, 3-GPU support.
v0.3.0 — Refonte complète : proxy LocalAI (plus de dépendance llama-cpp-2), bridge MCP (5 outils), détection multi-GPU, streaming SSE. Déployé sur EUREKAI:8769.# Build
cargo build --release
# Run
./target/release/hermes-rust-backend
# → Listening on 0.0.0.0:8769
# → Auto-connecte à LocalAI (192.168.1.47:8080)
# → 8 modèles disponibleshermes-rust-backend/
├── src/
│ ├── main.rs — Serveur Axum, auto-découverte LocalAI
│ ├── gguf_engine.rs — Proxy LocalAI (8 modèles, inférence réelle)
│ ├── openai_api.rs — /v1/chat/completions, /v1/models (streaming)
│ ├── api_server.rs — /status, endpoints Hermes
│ ├── mcp_bridge.rs — 5 outils MCP pour Hermes Agent
│ ├── hailo_engine.rs — Hailo-8 : classify/detect/OCR via MCP bridge
│ ├── multi_gpu.rs — Détection 3 GPUs (2×RTX3060 + GTX1060)
│ ├── auth.rs — JWT authentication
│ └── onnx_engine.rs — ONNX Runtime (détection automatique)
└── Cargo.toml| Endpoint | Method | Description |
|---|---|---|
/health | GET | Status + version + features |
/engines | GET | Tous les engines (GGUF, Hailo, CUDA, ONNX) |
/v1/chat/completions | POST | OpenAI-compatible (proxy LocalAI) |
/v1/models | GET | Modèles disponibles depuis LocalAI |
/status | GET | Status engines + GPU count |
/mcp/tools | GET | 5 outils MCP pour Hermes Agent |
mcp_gguf_infer — Générer du texte via LocalAI
mcp_hailo_classify — Classifier une image via Hailo-8 ResNet
mcp_hailo_detect — Détection YOLOv8m via Hailo-8
mcp_hailo_ocr — OCR via Hailo-8 Tesseract
mcp_engine_status — Statut de tous les engines# Sur EUREKAI (192.168.1.47):
cd ~/hermes-rust-backend
git pull
cargo build --release
pkill -f hermes-rust-backend
nohup ./target/release/hermes-rust-backend > /tmp/rust_backend.log 2>&1 &
# Vérifier
curl localhost:8769/health
curl localhost:8769/engines
curl localhost:8769/mcp/tools# Chat via proxy LocalAI
curl -X POST localhost:8769/v1/chat/completions \
-H "Content-Type: application/json" \
-d '{"model":"gpt-4o","messages":[{"role":"user","content":"Hello"}]}'
# MCP tools
curl localhost:8769/mcp/toolsllama-cpp-2 / libclang)~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.