agentleFS
ChanMeng666ChanMeng666/archlang/skills/archlang-review-plan/SKILL.md

archlang-review-plan

Skill33 starsChanged 21 days agoMITView on GitHub

Use when the user wants an existing ArchLang .arch floor plan checked rather than changed, e.g. "review this plan", "does this layout meet the brief?", "is every room reachable?", "check it for accessibility", "why does lint complain?". Reads the plan's facts, runs the lint profiles and the intent gate through the archlang CLI, and reports findings with suggested fixes. It edits nothing unless asked.

  • Installs packages

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 2 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    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 11
  2. Very specific:

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

    line 12
  3. Distinctive:

    Check against the brief, if there is one: Write the brief's checkable requirements as an intent file: room count only if the brief lists …

    line 34
  4. Distinctive:

    Do not apply them.

    line 46

≈ “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

Review a plan

Report what is true of a .arch plan and what is wrong with it, from the compiler's own facts. 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. Run archlang spec for the language, archlang context --section errors for the diagnostic catalog, and archlang <cmd> --help for flags.

The checks are drafting heuristics. They do not establish that a plan meets a building code, and your report must not say that it does.

Steps

  1. Does it compile? archlang validate plan.arch --json. Errors come first in the report; a plan with errors has no facts to review until they are fixed.
  2. Read the facts. archlang describe plan.arch --json gives rooms with areas and uses, what each door and window connects, the access graph and the totals. On a large plan narrow it: archlang describe plan.arch --select rooms,access,totals --json. area_m2 is measured on wall centrelines, so it counts the wall bands that stand on the room; net_area_m2 subtracts them. Say which one you quote.
  3. Run the default lint. archlang lint plan.arch --json. Each warning has a code, a place in the source and a fix in prose. archlang explain <CODE> --json explains a code you do not know.
  4. Run the accessibility profile when the user asks about access, or when the brief mentions a wheelchair, step-free access or ageing in place: archlang lint plan.arch --profile accessibility-advisory --json.
  5. Check against the brief, if there is one. Write the brief's checkable requirements as an intent file: room count only if the brief lists the rooms, an area band only where the brief gives a number. The format is the JSON Schema at https://archlang.uk/intent.schema.json, explained at https://archlang.uk/intent; a malformed file is refused with exit 3 and the path of the bad field. Then run archlang validate plan.arch --intent intent.json --feedback --json. A failed gating assertion exits 2 and comes with a correction prompt for each violation. archlang score plan.arch --brief intent.json --json gives the same comparison as a number that never fails.
  6. Ask for candidate fixes. For an unreachable room, a building with no entrance, a bathroom reached only through a bedroom, or a bedroom with no window, archlang suggest plan.arch --json returns door and window statements with a reason for each. Put them in the report as options. Do not apply them.
  7. Look at it, if that helps. archlang preview plan.arch --ascii prints a text plan; archlang preview plan.arch -o plan.png renders an image. archlang compile plan.arch --overlay circulation -o walk.svg draws the walking routes and their narrow points.

Report to the user

Order the findings so the user can act on them:

  1. Errors: the plan does not compile, with each diagnostic.
  2. Brief: each requirement the plan misses, with the measured value beside the asked value.
  3. Soundness: each lint warning, grouped by room, with the fix the diagnostic proposes.
  4. What you did not check, for example "no brief was given, so nothing was compared to one".

State which lint profile produced each warning. Do not call a warning-free plan compliant, safe or approved: say that the checks you ran raised nothing.

When something goes wrong

  • --code and --severity on archlang lint and archlang validate only filter what is shown. ok and the exit code still count every diagnostic, so a filtered result is not a clean one.
  • Exit code 3 is a wrong flag: run archlang <cmd> --help.

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.