SkillHub

Skills brillout skills

human-parsable

Use it when:
You want the agent's long messages, such as summaries of its work, reports, plans or PR descriptions, readable at a glance.
It does:
Gives the gist first, such as a TL;DR of bullet points with one section per bullet, and moves the details into collapsibles, a gist or a file.
Decisions:
Numbered, with lettered choices and the agent's recommendation highlighted, so you can reply "1B".
Unlike caveman:
That one makes every reply terse; this one leaves short answers alone and structures only the long ones.
Runs:
By itself, before the agent writes a long message.

TL;DR

It sets one goal for long messages: you get the gist in seconds, and can find any detail without reading everything. Short answers stay short, since structure there is overhead. The rest is a toolbox the agent picks from: a TL;DR of bullet points, then one section per bullet under the same name; details moved into collapsibles, a GitHub gist or a file; results ordered by what matters to you, not by the order the work happened; bold lead words, so your eye can scan the left edge; tables to compare items; before => after for changes; and a short snippet where it beats a paragraph. Decisions get numbered, with lettered choices and the agent's recommendation highlighted, so you can answer with something like "1B". The agent reads it before writing a summary of its work, a report, a plan, a review or a PR description, and at the start of a task that will end with one.

Summarizes the skill at 5f3a481.

npx skills add https://github.com/brillout/skills --skill human-parsable

skills.sh pageSource on GitHubAll skills from brillout/skills

Total Downloads
1 on skills.sh, Nov 9, 2026
First seen
Nov 6, 2026 on skills.sh
Weekly Downloads
1
Week to Nov 9
Weekly values
Week toInstalls
Sep 210
Sep 280
Oct 50
Oct 120
Oct 190
Oct 260
Nov 20
Nov 91
Category
category:output-stylecategory:software-engineering

Change how the agent replies: shorter, clearer, fewer tokens.

Everything for building software: planning, coding, debugging, design, frameworks, cloud, agents and skills.
Type
type:guidelines

Knowledge and rules the agent applies while it works: best practices, design principles, API references.
Effect
effect:none

Adds no side effects of its own. It shapes how the agent thinks, writes or reviews. Guidelines and routers are always effect:none, even when the task they support edits code.
Needs
needs:none

Instructions only. Works in any agent that supports skills.
Publisher
by:community

Everyone else. Unverified vendor claims default to community.
Topics
output-style

SKILL.md

From brillout/skills at 5f3a481. Checked daily, last changed Nov 9, 2026.

Frontmatter
name: human-parsable
description: >-
  Efficiently communicating large amounts of information to humans.

  For example: a final summary of your work, a report, a plan, a review, a PR description.
  Before writing such a message, you must always read this skill.
  For a task that will end with such a message, read it when you start the task.

Goal

Make large amounts of information human-parsable: the human gets the gist in seconds and finds any detail without reading everything.

That's the only rule. Everything below is a toolbox of examples — use what helps, skip what doesn't, be creative (don't limit yourself to the examples below).

Small amount of information => just answer, structure is overhead.

Examples

  • TL;DR — as bullet points, everything important at a glance
  • Sections — one ## Subsection per TL;DR bullet, same name and same order (skip it if the bullet says it all)
  • Bury details, e.g. in collapsibles, GitHub gist, file, ...
  • Decisions
    • List and enumerate decisions => the human sees all decisions at a glance
    • Highlight the choices and your recommendation
    • Make choosing easy, e.g. numbers for decisions and letters for choices => the human can reply 1B
  • Highlight
    • What the human must know
    • Results, not the journey — order by importance to the human, not by the order you did things
  • Bold lead words => the eye scans the left edge
  • Tables to compare items across the same attributes
  • Before => after, e.g. timeout: 5s => 30s
  • Show, don't tell, e.g. a 3-line snippet might beat a paragraph