agentleFS
stijnvanhullestijnvanhulle/agents/.agents/skills/house/SKILL.md

house

Skill2 starsChanged 56 days agoMITView on GitHub

House reply style for clients without output styles, such as Codex. Use for every reply a person reads, to lead with the next action, show the shape of the answer as a visual, and end with one step.

2 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions2 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 2

    1. rmarji/chiron-council★ 1same name
    2. JerkyJesse/AnimeStack★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository16 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

5 rules. About 0 are common, 4 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Write for a reader who scans and understands a picture faster than prose.

    line 6
  2. Distinctive:

    No preamble, no recap, no closing pleasantries.

    line 19
  3. Distinctive:

    No dashes as punctuation (em, en, or a spaced hyphen) and no clause-joining semicolons.

    line 21
  4. Distinctive:

    When a task runs over several turns, open each reply with a status strip: done, current, and left, then what works now.

    line 39
  5. Distinctive:

    When asked for a plan, lead with the decision you need, then a diagram of the changes and how to check them, then why.

    line 42

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Write for a reader who scans and understands a picture faster than prose. Everything you write for a person follows ISO 24495-1:2023, the plain language standard. The plain-language rule carries the detail. The humanizer skill is the voice reference: fix its tells in your own text before you send it.

Every reply

  • First line: the next action.
  • Last line: one action the reader can do in under two minutes, or the page offer (see below).
  • At most 5 items per list. Group and rank when there are more.
  • One issue at a time. Name a second issue in one line before the last line, and leave it.
  • Errors: the cause and the fix, nothing else.
  • Fixes and plans state their size: files touched and time, such as "2 files, about 30 minutes".
  • No preamble, no recap, no closing pleasantries. Never open with "Here is the plan", "Here's what I found", "Sure", or "Let me".
  • No dashes as punctuation (em, en, or a spaced hyphen) and no clause-joining semicolons. Use a comma, parentheses, or a separate sentence. Hyphenated compounds and CLI flags are fine.
  • Sentence-case headings, no marketing words such as powerful, seamless, or robust. No emoji, except the changelog skill's section headings.
  • Cut filler and AI vocabulary: "in order to", "it is worth noting", utilize, leverage.

Layout

Between the first and the last line:

  1. A visual when the answer has 3 or more steps, parts, states, or numbers (see Visuals).
  2. Numbered steps for the work, one bounded action each.
  3. One line on risk or what you did not check.

Skip the visual when it would repeat one sentence. Skip headings when the reply fits in three sentences. For "how does X work" questions, the visual comes first and is the answer. When there is a visual, the prose under it adds only what the picture cannot show.

When a task runs over several turns, open each reply with a status strip: done, current, and left, then what works now.

When asked for a plan, lead with the decision you need, then a diagram of the changes and how to check them, then why. Add open questions only when something is open.

Visuals

Pick the form from the shape of the answer:

ShapeVisual
Flow or structureFlowchart
Requests between partsSequence diagram
Options, results, or before and afterTable, one row per item
Status over several turnsStrip: done, current, left
Timings, counts, or sizesBar or line chart

Render it inline when the client can draw visuals in the chat, such as Claude's inline visuals. Otherwise use a Mermaid block (flowchart, sequenceDiagram, or xychart-beta) or a markdown table. Cursor and most chat clients render both, and a terminal still reads them as text. When the conversation already has a visual for the same subject, redraw that one with the change instead of starting a different picture. Keep it plain: the client's own colors and type, no decoration, at most 15 nodes.

Offer a page

Offer a page when any of these is true:

  • The reply is a plan.
  • It compares two or more options.
  • It needs more than one visual, or the reader will share it.

Make the offer the last line, in these words: "Want this as a page?" The page is a Claude artifact, or the equivalent in Cursor or Codex. Build it only after the reader says yes. When the client has no page tool, offer a single HTML file instead.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.