Make Hermes your own.

theme-only desktop plugin: it registers eighteen themes through the plugin SDK, and their stylesheets restyle parts of Desktop's own UI (sidebar art, tabs, composer, wordmark) only while a Pantheon theme is active. Fonts and art are bundled; no network requests, no agent tools or hooks.

probe.py re-executes itself under the Hermes Python interpreter located via HERMES_PYTHON / VIRTUAL_ENV / the parent process or the Hermes installer's managed venv (HERMES_HOME/installs); it reads provider API keys from the Hermes .env, Hermes's own saved OAuth logins (auth.json credential pool, read-only), the Vaultwarden plugin's local secret cache (cache/vaultwarden_cache.json, read-only, only when Vaultwarden is the sole enabled secret source) and Docker s6 container env files, plus other coding-assistant CLIs' local logins (including macOS keychain entries), and sends each key only to its own vendor's usage endpoint.

feeds and full articles are fetched with curl through the gateway's shell.exec on the connected Hermes host (public-IPv4 SSRF guard); opt-in AI grading sends article excerpts to your configured model and creates HERMES_HOME/skills/rss-importance-grading/SKILL.md if missing; opt-in image loading fetches publisher images directly from Desktop.

the gateway listens for devices on a WebSocket port (8765, all interfaces by default; plain ws:// unless tls_cert/tls_key are set). New devices need a Hermes pairing code unless listed in GADGET_ALLOWED_USERS; GADGET_ALLOW_ALL_USERS skips pairing (development only). A paired device answers the agent's command approvals for its own chat (allow once or deny). The first approved device becomes the platform's home channel by default. Spoken replies that are not WAV are converted with the local ffmpeg. `hermes gadget update --latest` downloads firmware from this repository's GitHub releases, checks size and SHA-256 against the release manifest (files carry build provenance attestations) and sends it to the device. Writes only under plugin-data/gadget/. Uses Hermes's own STT, TTS and image cache; no telemetry. Independent community project, not affiliated with Nous Research.

spawns the vendor devin CLI from PATH or cli_path (devin acp, devin acp --cloud, devin --print, devin auth status, devin models list); the CLI edits files and runs shell commands in a folder the model picks (any folder unless allowed_roots is set) under the configured permission mode (accept-edits/smart auto-allow edits, dangerous auto-allows everything and needs allow_dangerous: true), and talks to Cognition's servers; Devin Cloud runs use Cognition-managed permissions that Hermes does not gate. Every model-initiated launch goes through Hermes approval; follow-ups typed on a Desktop run card and answers to Devin's prompts on the card start or allow actions on that click without a Hermes prompt (unanswered card prompts are denied after 300 s by default). Uses the existing Devin CLI login and stores no credentials; Hermes provider, bot and tool secrets are stripped from the child environment (extra_env is passed as set). Background runs continue after the turn up to their timeout and post Devin's output back into the conversation as a message when they finish. Run history (prompts, diffs, command output) and model preferences are saved under plugin-data/devin unless persist_runs is off. Independent community add-on, not affiliated with Cognition.

sends queries, result limits, gateway IDs, and optional BYOK aliases to api.cloudflare.com, which forwards searches to the selected provider and records them in AI Gateway logs. Requests can consume gateway credits or provider quota. Reads its own settings and CLOUDFLARE_WEB_SEARCH_API_TOKEN through Hermes' profile-scoped APIs. Registers one web provider, with no new tools, hooks, background processes, file writes, or self-updater. Independent community plugin.