Nous Research Ships Bot Mode for Hermes Agent

Bot Mode first appeared as a one-day public beta plugin from co-founder Teknium, who asked for bug reports and committed to folding it into the main desktop app. That integration has now happened. According to the company, the standalone repository is archived and development continues in-tree under apps/desktop/src/plugins/hermes-bots/.

The core idea is that a bot is just a profile stored under ~/.hermes/profiles/<name>, with isolated config, memory, skills, credentials, and chat history. The plugin is a user interface over a primitive Hermes already shipped, which keeps the surface area small. Profile creation and editing use the profiles.* gateway RPCs, avatar generation uses image.generate, and routines run as ordinary Hermes cron jobs namespaced [bot:].

Bots communicate through a persistent Agent Inbox conversation. Messages between bots are real command-line handoffs, arriving with attribution, and each bot’s SOUL.md teaches the reply protocol. Typing an @mention hands off work to another bot and reports back.

The New Agent dialog asks for name, title, and description, with an advanced section opening the full profile configuration. Users can clone an existing profile, pin a provider and model, write a custom SOUL.md, or skip skills. Right-clicking a bot exposes Edit Profile, covering per-skill and per-toolset enablement, the model pin, and SOUL.md. Duplicates clone config, skills, memory, and appearance. Avatars are geometric faces in seven shapes and ten colors, with options to upload an image, generate a portrait, or attach a pixel pet.

The in-tree version also added capabilities beyond the beta. Groups organize the roster into labeled sections that sync across machines. Group chats open a shared room for two to six bots, with up to three serial rounds of member turns; mentioned bots respond, and when nobody is mentioned, everyone responds briefly or passes. A multi-source roster pulls bots from every connection under Settings → Connections.

Nous Research positioned the feature for solo builders, startups, and small-to-mid engineering teams, which can adopt it immediately. Enterprises should treat it as a workstation tool, not managed infrastructure, since there is no admin console, SSO, central audit log, or policy layer. Regulated buyers will need their own controls around it. The company listed software engineering, AI research labs, technical media and content operations, developer relations, quantitative research, and IT operations as likely application areas.