Next Prompt — Hermes Plugin

AI-generated follow-up prompt suggestions for Hermes Agent Desktop.

After each completed turn, Hermes quietly asks the model whether there is a natural next step and, if there is, shows it as a small pill under the composer. Click it to use the text — it is never sent for you. Click × to dismiss it.

On Hermes Desktop builds with the plugin SDK's composer API (host.composer.insertText), the click puts the text straight into the composer. On earlier builds it copies the text to the clipboard and tells you so; paste it with Ctrl/⌘+V.

Install

hermes plugins install next-prompt        # once listed in the plugin catalog
# or straight from GitHub:
hermes plugins install soymarketing/hermes-plugin-next-prompt
hermes plugins enable next-prompt

Then restart Hermes Desktop and turn Next Prompt on in Settings → Plugins (desktop plugins are opt-in).

How it works

post_llm_call hook — when a turn finishes in a Desktop/TUI session, a background worker asks ctx.llm for one short follow-up (or NULL when there is nothing useful to suggest).
Stays until you act on it — the suggestion is kept per chat in the plugin's data directory (ctx.state.data_dir, under the profile's plugin-data/), so switching screens or chats, or a backend restart, does not lose it. It goes away when you use it, dismiss it, or send your own next message (pre_llm_call). Stale suggestions expire after 7 days.
Desktop half — reads the suggestion through the plugin's REST route (ctx.rest → /api/plugins/next-prompt/suggestion) and renders it in the composer.underside area.

Messaging platforms (Telegram, Discord, …) are skipped: nobody would see the pill there, so no LLM call is spent on them.

Configuration

Suggestions run on your main model by default. To use a cheaper or faster model, point the plugin's auxiliary task at it in config.yaml:

auxiliary:
  next_prompt:
    provider: openrouter
    model: vendor/small-fast-model

Plugin settings:

plugins:
  entries:
    next-prompt:
      settings:
        enabled: true            # toggle suggestions on/off
        max_context_messages: 4  # recent messages sent for context (2-10)
        cooldown_seconds: 3      # minimum gap between suggestions per chat

Privacy and cost

The last few messages of the chat (each truncated to 500 characters) are sent to the configured model once per completed turn.
One short completion per turn (max_tokens=100), on the model you choose.
No telemetry, no network access other than the host's own ctx.llm call.

Requirements

Hermes Agent ≥ 0.21 with the Desktop app.

Design notes

Pill, not ghost text. The plugin SDK does not expose the composer's input, so the suggestion lives in composer.underside, a first-class extension area.
Insert, don't send. The click only ever places the text for you to review. The bundle stays inside the plugin SDK: it uses host.composer.insertText when the host has it and falls back to the clipboard otherwise — it never touches the app's composer DOM or internal events, so composer changes can't break it silently.
Poll, not push. ctx.socket is a no-op on OAuth remote backends, so the desktop half polls (every 2 s right after a turn, every 20 s otherwise).
No suggestion beats a bad one. The model is told to answer NULL when the task is done or the agent asked you a question.

Development

git clone https://github.com/soymarketing/hermes-plugin-next-prompt
cp -r hermes-plugin-next-prompt "$HERMES_HOME/plugins/next-prompt"
hermes plugins validate "$HERMES_HOME/plugins/next-prompt"
hermes plugins enable next-prompt

Python changes need a Desktop restart; desktop/plugin.js hot-reloads.

License

MIT

next-prompt

What it adds

Hooks 2

README