hermes-skills

Detect local GPU hardware and recommend optimal llama.cpp model + KV cache configuration. Supports AMD (ROCm) and NVIDIA (CUDA). Recommends TurboQuant KV compression when standard f16 cache won't fit.

Email management via himalaya CLI with Bitwarden-backed authentication.

Set up and use Bitwarden Secrets Manager CLI (bws). Use when reading, listing, or injecting secrets into env vars and dotenv files.