Hermes Multichannel Prompt Optimizer

What Hermes Multichannel Prompt Optimizer does

Hermes Multichannel Prompt Optimizer was a Hermes Agent plugin that rewrote prompts before they reached the model, across the CLI, TUI, Discord and Telegram, with quality scoring, analytics and arrow-key interactive overlays. The author retired it on 3 October 2026 and archived the repository, so it is read-only and no longer maintained.

The retirement notice explains why. In a controlled test on 100 real WildChat-1M messages, each answered once as written and once after rewriting, with a judge comparing the pairs in both orders, neither tested version produced answers that were preferred on average. The development version rewrote 67 of 100 prompts and the rebuilt version rewrote 19. Earlier claims of better answers and a gain of 55 quality points per rewrite were withdrawn. The code remains under the MIT licence, and the README gives commands to disable and remove the plugin.

Key features

• Rewrote prompts before they reached the LLM
• Worked across the CLI, TUI, Discord and Telegram
• Prompt quality scoring and analytics
• Arrow-key interactive overlays
• Local metrics.db log of original and rewritten prompts, stored only on your machine

When to use it

• Read the published test results before building a prompt-rewriting layer of your own
• Remove the plugin cleanly if you installed it earlier
• Study the archived code as an example of a multichannel Hermes plugin

Who it is for: Hermes Agent users who installed the plugin and need to remove it, and developers studying a tested prompt-rewriting approach.

How it fits with Hermes Agent

It is a plugin built for Hermes Agent, managed through the hermes plugins commands, and it is now retired.

Note: The repository is archived and read-only, and the author retired the plugin on 3 October 2026 after a test showed no improvement in answers.

FAQ