Fix: remove empty extra: line in config.yaml

Fix: remove broken hermes.dnsxi.art domain, add API_SERVER_KEY

Add log proxy in nginx

Add log exposure for debugging

Revert entrypoint to original

Add pm2 log output for debugging

fix: install python-telegram-bot separately to avoid build error

fix: use single quotes for pip install messaging extra

fix: install hermes-agent[messaging] for python-telegram-bot support

fix: always generate fresh config.yaml (backup data not in config)

fix: always set telegram base_url via CLI (backup may restore old config)

feat: use cf-worker-telegram as base_url for native TG integration

feat: native telegram platform via tg.dnsxi.art proxy, remove web-ui dependency

fix: rclone backend changed to hi168 S3, use sync.sh for restore on startup, initial backup after config

fix: remove web-ui, nginx proxies directly to API server 8642, API_SERVER_HOST=0.0.0.0

fix: use hermes config set CLI to set model credentials from env vars

fix: remove base_url/api_key from config.yaml - let .env take precedence

fix: add aiohttp dependency for API server platform

fix: add api_server platform to default config.yaml in entrypoint

fix: don't remove api_server from config.yaml - let web-ui manage it

fix: force-remove api_server from config.yaml to use loopback no-auth mode

fix: remove API_SERVER_KEY - use loopback with no auth, web-ui proxies to API server

fix: revert nginx to single upstream - web-ui proxies /v1/ internally

fix: API_SERVER_HOST=0.0.0.0 for nginx access, pm2 --update-env

fix: proxy /v1/ to API server port 8642, not web-ui 8648

fix: embed api_server key in config.yaml, properly export env vars for pm2

fix: use double-quote heredoc for .env variable expansion

feat: configure Hermes with LLM provider + API server key on startup

feat: add cron auto-backup every hour + install cron in Dockerfile

deploy Hermes agent with rclone backup

initial commit