Auxiliary model switching now looks up provider details from cfg
registry (base_url, api_key) and adds custom: prefix automatically,
matching the behavior of main provider switching.
— 小糯
- providers/list/test now read from cfg (HERMES_CUSTOM_PROVIDERS)
- switch writes target provider into config.yaml custom_providers and restarts gateway
- standard providers (openrouter, anthropic) clear custom_providers on switch
- fallback to config.yaml if cfg unavailable
Signed-off-by: Xiaonuo <xiaonuo@shazhou.work>
Supports api_key_env field in custom_providers config, reads
from process.env first then falls back to ~/.hermes/.env.
Signed-off-by: Xiaonuo <xiaonuo@shazhou.work>
- models providers: list configured custom providers
- models list [provider]: fetch and display available models from /v1/models
- models test [provider]: connectivity check with latency
- models status: show current main + auxiliary model config
- models switch: change provider/model for main or auxiliary tasks
- --aux <task>: target specific auxiliary tasks (repeatable)
- --tg: switch via Telegram /model command (current session)
- --global: persistent switch (with --tg)
Signed-off-by: Xiaonuo <xiaonuo@shazhou.work>
CRUD for Hermes personality presets in ~/.hermes/config.yaml.
Supports --tg flag to switch via Telegram Bot API for live session updates.
Usage:
hermes-har personality list
hermes-har personality show <name>
hermes-har personality add <name> <prompt>
hermes-har personality remove <name>
hermes-har personality switch <name> [--tg]
