nopus plugin
A plugin that measures language complexity in coding-agent replies and sends an overly abstract reply back for a rewrite
- 299 ★ on the repository
- Last commit
Extended access
Why this level
- The Stop hook runs a bundled Node.js script after every agent reply
- Each rewrite adds a model request and spends tokens
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add plugins/nopusDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The Stop hook runs a bundled Node.js script after every agent reply
- Each rewrite adds a model request and spends tokens
This is third-party code. Review the repository files before installing.
What it does
nopus triggers when the agent finishes a reply. The plugin strips code blocks, inline code, URLs, file paths and table rows from the text. It then measures the share of rare words, abstract vocabulary, noun stacks and dense phrases. Word frequency comes from bundled data on conversational and broad web text, abstractness from human concreteness ratings, and a computing glossary softens the weight of technical terms. The plugin decides on a combination of measures, and a single rare word triggers nothing. If a reply crosses the threshold of the chosen sensitivity, the agent gets a request to rewrite it more plainly while keeping the meaning and necessary detail. The request carries examples from the reply itself, such as the stock phrase "here's where it gets interesting". The plugin asks for a rewrite at most once, so there is no retry loop.
Who it is for. For people who read many coding-agent replies and are tired of abstract phrasing such as "capability ownership strategy" in place of concrete actions.
Good fit when
- The agent replies in long abstract phrases and you want more concrete answers without editing every prompt
- You want to choose how often it steps in: the low, medium and high levels gave roughly 5.3, 9.9 and 18.6 percent rewrites on the Pi sample
- You want a short first answer: extra-simple mode keeps the conclusion, the next action and important warnings
- One reply felt heavy while the plugin stayed silent: the nopus-simplify skill rewrites the previous reply on your command
- You need the same check in Pi, Claude Code and Codex with a shared settings file
Not a fit when
- Texts in Russian: the examples, frequency dictionaries and stock-phrase list belong to English, and the author does not promise it works on Russian replies
- You need an article or document edited: the plugin looks only at agent replies
- You cannot spend extra tokens: every rewrite means one more model request
- You need your company style rules: the set of measures and stock phrases is fixed
Example request
Set nopus sensitivity to high and turn on extra-simple mode: I want a short first answer with the conclusion and the next action.Limitations
Node.js 22 or newer with the `node` command on PATH is required. The plugin supports Pi, Claude Code and Codex. In Pi the reply is visible while streaming and is hidden only after a rewrite request, and the rejected reply stays in the model history. The stock-phrase list is a short set of literal phrases, so it does not catch every sign of machine style. The author computed the intervention rates on 5,337 completed Pi replies, and other agents and tasks will produce other values. Settings live in `$XDG_CONFIG_HOME/nopus/config.json`, and `NOPUS_*` variables override them.
How to disable. Return replies to normal with `/nopus off` in Pi or through an environment variable. To remove the plugin entirely, use `/plugin` in Claude Code and `codex plugin` in Codex. The `config.json` file in the settings directory stays; delete it by hand.
FAQ
How do I install nopus in Claude Code?
Run `/plugin marketplace add Vistyy/nopus` and `/plugin install nopus@nopus`. The identifier follows the plugin@marketplace scheme, so here it is `nopus@nopus`. The plugin uses a Stop hook.
How does nopus decide a reply is too complex?
It counts rare words, abstract vocabulary, noun stacks, phrase density and stock phrases such as "here's where it gets interesting". A combination of measures is needed; one rare word does not trigger a rewrite.
Can the plugin trap the agent in a loop?
No, the plugin asks for an automatic rewrite at most once per reply. The host shows the message "nopus requested a clearer rewrite."
How do I change the sensitivity?
Pick low, medium or high. In Claude Code use `/nopus:nopus-configure medium`, in Codex `$nopus-configure medium`, in Pi `/skill:nopus-configure medium`. The default is medium.
What does the nopus-simplify skill do?
It rewrites the last agent reply more plainly, keeping the meaning, terms, commands, paths, errors and conditions. Only you run it; the model does not call it by itself.
Discussion
No questions or reviews yet.
Related
PluginsJob search inside your coding agent: scores each listing against your CV from 1 to 5, tailors a PDF resume, drafts letters and tracks applications
A Claude Code workflow where the agent searches job postings, scores fit against your profile, drafts a CV and cover letter, and helps prepare for interviews
A skill that makes the agent open with the action, number the steps and skip tangents
Obsidian skills
Obsidian Skills
Skills for agents working with Obsidian: Markdown with wikilinks, Bases, JSON Canvas, Obsidian CLI, Defuddle and Knap
A self-hosted knowledge base with block-level references and a built-in MCP server for connecting AI agents to your notes
The official Reactive Resume MCP server and skill: the agent builds and edits resumes and runs a job application tracker
Comparison with similar tools
Entries that share the main search query, and how they differ.
README in short
The author shows a real agent reply, the problems found in it and the rewritten version. A separate evaluation page explains where the intervention rates come from.