concise

concise

say more with less

Install • Before/After • Elaborate • vs Caveman • Why

---

A Claude Code skill/plugin and Codex plugin that makes Claude respond concisely -- cutting ~60-70% of output tokens while keeping natural, readable English. Also works with Cursor, Cline, Windsurf, and GitHub Copilot.

Reduce token usage. Lower API costs. Get faster responses. No broken grammar. No telegram-speak. Just the answer, tight and clean.

Before / After

Normal Claude (69 tokens)

"The reason your React component is re-rendering is likely because you're creating a new object reference on each render cycle. When you pass an inline object as a prop, React's shallow comparison sees it as a different object every time, which triggers a re-render. I'd recommend using useMemo to memoize the object."

Concise Claude (27 tokens)

"You're creating a new object reference each render. Inline object props fail shallow comparison every time. Wrap it in useMemo."

Normal Claude

"Sure! I'd be happy to help you with that. The issue you're experiencing is most likely caused by your authentication middleware not properly validating the token expiry. Let me take a look and suggest a fix."

Concise Claude

"Bug's in auth middleware -- token expiry check uses < instead of <=. Here's the fix:"

Normal Claude

"Sure! Merge conflicts happen when Git can't automatically reconcile differences between two branches. Let me walk you through the process of resolving this. First, you'll want to open the file and look for the conflict markers..."

Concise Claude

"Open the conflicted file, find the <<<<<<< markers. Keep the code you want from each side, delete the markers. Then git add and commit."

Same answer. Proper English. Fewer tokens.

Elaborate on Request

The key feature: ask for more detail anytime and get it. Then concise auto-resumes.

You:     Why is my app slow?
Claude:  Main bottleneck is N+1 queries in the orders endpoint. Each order
         triggers a separate DB call for line items. Use includes(:line_items)
         to eager-load.

You:     Elaborate on that
Claude:  When Rails loads your orders list, it runs one query to fetch all
         orders. Then for each order, when you access order.line_items in the
         view, ActiveRecord fires a separate SELECT query. With 50 orders,
         that's 51 total queries -- one for the list, plus one per order.
         This is the N+1 query problem.

         includes(:line_items) tells ActiveRecord to load all line items in
         a single additional query. Instead of 51 queries, you get 2. On a
         typical page, this cuts response time from seconds to milliseconds.

You:     OK, what about the frontend?
Claude:  Frontend's fine. Lighthouse shows 92 performance. Only issue is an
         unoptimized hero image -- run it through ImageOptim.

No permanent mode switch. One detailed answer, then back to concise.

Install

npx skills add o4f6bgpac3/concise

For a specific agent:

npx skills add o4f6bgpac3/concise -a cursor
npx skills add o4f6bgpac3/concise -a github-copilot
npx skills add o4f6bgpac3/concise -a cline
npx skills add o4f6bgpac3/concise -a windsurf
npx skills add o4f6bgpac3/concise -a codex

Or with Claude Code plugin system:

claude plugin marketplace add o4f6bgpac3/concise
claude plugin install concise@concise

Usage

Trigger with:

• /concise
• "concise mode"
• "be concise"
• "less verbose"

Stop with: "stop concise" or "normal mode" or "be verbose"

What Concise Does

Thing	Concise does...
English explanations	Cut filler, keep grammar
Code blocks	Write normally
Technical terms	Keep exact
Error messages	Quote exact
Git commits & PRs	Write normally
Articles (a, an, the)	Drop where natural, keep where needed
Pleasantries	Gone ("Sure! I'd be happy to..." is dead)
Hedging	Gone ("It might be worth considering..." is extinct)
Elaborate requests	Full detailed answer, then auto-return

vs Caveman

Caveman and concise are complementary:

	Caveman	Concise
Token savings	~75%	~60-70%
Grammar	Broken (fragments, no articles)	Correct (natural English)
Readability	Requires adjustment	Reads like normal text
Intensity levels	lite / full / ultra	Single mode
Elaborate on request	No	Yes
Best for	Maximum compression	Balanced readability + savings

Want maximum compression and don't mind telegram-style? Use caveman. Want readable output with solid token savings? Use concise.

Concise was inspired by caveman -- loved the idea, wanted something that reads like normal English.

Why

• Faster responses -- fewer tokens to generate means faster output
• Easier to read -- no walls of text, just the answer
• Same accuracy -- all technical info preserved, only filler removed
• Lower cost -- ~60-70% fewer output tokens
• Still sounds human -- proper English, not broken grammar

Development

git clone https://github.com/o4f6bgpac3/concise.git
cd concise
git config core.hooksPath .githooks

Edit only skills/concise/SKILL.md. The pre-commit hook syncs all other copies automatically.

See CONTRIBUTING.md for details.

License

MIT