agentleFS
ChanMeng666ChanMeng666/archlang/skills/archlang-brief-to-plan/SKILL.md

archlang-brief-to-plan

Skill33 starsChanged 21 days agoMITView on GitHub

Use when the user gives a written brief for a new floor plan (rooms, areas, how rooms connect, a plot size) and wants the drawing, e.g. "design a two-bedroom flat of about 75 m²" or "lay out a small office with four rooms". Writes an ArchLang .arch file from the brief, checks it against the brief with the archlang CLI, and finishes it as a sheet. Not for changing an existing plan (archlang-edit-plan) or for exporting one (archlang-export).

  • Installs packages

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 3 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

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

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

    If archlang is not on the path, write npx -y @chanmeng666/archlang in its place.

    line 12
  2. Very specific:

    Never run a bare arch: on Linux, macOS and Git Bash that name is the coreutils command.

    line 13
  3. Very specific:

    Run the archlang CLI commands below and no other program.

    line 19
  4. Very specific:

    Write one new plan file, plan.arch here or the name the user gives, in the user's working directory, plus the drawings the CLI renders …

    line 20
  5. Very specific:

    Read the language: Run archlang spec and read all of it before writing a line.

    line 27
  6. Distinctive:

    List what the brief asks for: Write down each room, each stated area, each stated connection ("opens off the hall"), each stated window, …

    line 29
  7. Distinctive:

    Every requirement of the brief the plan does not meet, or "none".

    line 56
  8. Distinctive:

    Every lint warning still present, with the reason it stays.

    line 59

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

Brief to plan

Turn a written brief into a checked .arch floor plan. Every step is a CLI call; never judge the plan from the SVG markup. If archlang is not on the path, write npx -y @chanmeng666/archlang in its place. Never run a bare arch: on Linux, macOS and Git Bash that name is the coreutils command.

This skill is a workflow, not a language reference. The language is whatever archlang spec prints.

Scope

  • Run the archlang CLI commands below and no other program.
  • Write one new plan file, plan.arch here or the name the user gives, in the user's working directory, plus the drawings the CLI renders beside it. If a file of that name already exists, ask the user before replacing it.
  • The user's explicit instructions take priority over these steps.

Steps

  1. Read the language. Run archlang spec and read all of it before writing a line. For one command's flags and examples run archlang <cmd> --help, for example archlang finish --help.
  2. List what the brief asks for. Write down each room, each stated area, each stated connection ("opens off the hall"), each stated window, and any outer size. Keep numbers exactly as the brief states them. If two requirements cannot both hold, stop and tell the user which ones; do not quietly change a number.
  3. Write plan.arch. Give every room an id, a label and a uses. Attach doors and windows to walls and place furniture by anchor, as archlang spec shows, instead of computing coordinates. Give every room a door or an opening.
  4. Compile. archlang compile plan.arch -o plan.svg --json. If ok is false, read each diagnostics[].fix. Preview the mechanical fixes with archlang fix plan.arch --dry-run --json, then apply them with archlang fix plan.arch. Fix the rest by hand and compile again.
  5. Check the plan against your list. archlang describe plan.arch --select rooms,doors,windows,totals --json. Compare room count, labels, each area, what each door connects and which rooms have windows with the list from step 2. Edit and repeat from step 4 until they match. area_m2 is measured on wall centrelines; net_area_m2 is the floor no wall stands on.
  6. Check soundness. archlang lint plan.arch --json. Fix each warning, or keep it on purpose and say why. If a room cannot be reached or a bedroom has no window, archlang suggest plan.arch --json returns door and window statements you can paste.
  7. Finish the drawing. archlang finish plan.arch --dry-run --json shows the furniture and sheet statements it would add; archlang finish plan.arch applies them. Read unresolved: it names each room left empty and each statement left out, with the reason.
  8. Gate and show. archlang validate plan.arch --strict --json must exit 0, or you must be able to name each remaining warning as deliberate. Then archlang preview plan.arch -o plan.png to show the user.

Report to the user

  • The file you wrote and the drawing you rendered.
  • Every requirement of the brief the plan does not meet, or "none".
  • Which area you quote: net_area_m2 is the floor no wall stands on, area_m2 the larger centreline figure. If a room has no net_area_m2, say its net area could not be computed.
  • Every lint warning still present, with the reason it stays.

When something goes wrong

  • Exit code 2 is a fault in the source. Read the diagnostic and fix it; do not run the same command again unchanged. 3 is a wrong flag: run archlang <cmd> --help.
  • For an unfamiliar code, run archlang explain <CODE> --json.
  • To gate on the brief as data instead of by eye, write it as an intent file and run archlang validate plan.arch --intent intent.json --feedback --json. The archlang-review-plan skill covers this.

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.