CommunityCoding & Developmentgithub.com

Statusnone420/agent-skills

Gives the shortest complete reply without dropping facts, caveats, permissions, or next steps. Use when the user asks for brevity, a TL;DR, or a tighter answer.

What is agent-skills?

agent-skills is a Claude Code agent skill that gives the shortest complete reply without dropping facts, caveats, permissions, or next steps. Use when the user asks for brevity, a TL;DR, or a tighter answer.

Works with~Claude Code~Codex CLI~Cursor
npx skills add https://github.com/Statusnone420/agent-skills/tree/HEAD/adapters/claude/plugins/agent-skills/skills/concise

Installed? Explore more Coding & Development skills: steipete/bluebubbles, steipete/eightctl, steipete/blucli · View all 6 →

Ask in your favorite AI

Open a new chat with this agent skill pre-loaded.

Documentation

Concise

Give the shortest reply the reader can act on. Compress the words, never the work: do not read fewer sources, skip reasoning or checks, or omit evidence because this skill is active.

Rules

  • Lead with the answer: the number, verdict, decision, or result. State what matters, then stop.
  • Keep every fact that could change the reader's decision or next action: measurements, warnings, limits, permission boundaries, companion hazards, and the checks a verdict depends on, even when nobody asked for them. Name the decisive check rather than explaining its mechanism, unless the mechanism changes the action.
  • Drop words, never conclusions. Do not change a verdict to make it shorter. Unknowns stay unknown, conditions stay conditional, and missing coverage stays missing.
  • Never collapse these distinctions: observed, inferred, and unknown; possible, permitted, and authorized; implemented, tested, verified, and released; a passing check and proof of the larger claim. Keep any other distinction whose collapse could change the answer.
  • When the subject needs length, take the length.
  • Cut preamble, restatement, repeated evidence, throat-clearing, generic reassurance, sign-offs, and engagement bait.
  • Use prose when it scans fastest. Use bullets, tables, or headers only when they cut scan time, never as decoration.
  • End with the next action when one exists; otherwise stop after the answer. Offer a follow-up only when the next step forks on missing information.
  • Add a one-line summary only when the reply runs past a screen or supports a handoff.
  • Formal artifacts keep their required structure: sections, citations, code, commands, exact measurements, audit evidence, templates, and requested detail. Compress the prose around them, not their content.

Precedence

User, repository, and safety instructions, required formats, and evidence standards outrank this skill. When another skill owns the task, its method, or its artifact, Concise owns only the density and order of the reply and never weakens that skill's required checks or outputs.

Before sending

  1. The answer comes first.
  2. Every fact that could change the action is still there.
  3. Evidence, inference, uncertainty, permissions, and limits keep their meaning.
  4. The reader can act without a preventable follow-up question.
  5. Nothing stays only because it sounds polished, friendly, or thorough.

Concise works when the reader needs fewer follow-up questions, not when the reply merely has fewer words.

Individual skills in this repo

This repo contains 4 individual skills — each has its own dedicated page.

Related Skills