agentleFS
niksavisniksavis/basicly/.agents/skills/validate-as-consumer/SKILL.md

validate-as-consumer

Skill2 starsChanged 6 days agoMITView on GitHub

Runs a verified change the way a consumer does, against its requirement. Use at VALIDATE or before a README claims a capability.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    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. Distinctive:

    Read existing verification evidence.

    line 10
  2. Distinctive:

    Ask: could this transcript exist without the change?: If yes, you ran the suite again and validated nothing.

    line 14
  3. Distinctive:

    Look for a missing prerequisite, an error that names an internal path, or an instruction that assumes this repository.

    line 25
  4. Distinctive:

    When the change ships to a consumer, run it there: a new directory, an install, and the documented command.

    line 30
  5. Distinctive:

    Check each acceptance criterion against the picture.

    line 43
  6. Very specific:

    When the repository supplies an overflow checker, run it and read both of its signals, render-overflow and render-overlap.

    line 45

≈ “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
<!-- Generated by `basicly skills-build` from skill.yaml. Do not edit; edit the source. -->

Validate as a consumer

This skill checks a verified change as a consumer uses it. Read existing verification evidence; repeating deterministic checks alone is not consumer validation.

Rules

  • Ask: could this transcript exist without the change? If yes, you ran the suite again and validated nothing.
  • Check the whole task contract. Assess Context, every Acceptance criterion and any distinct Requirements together; the builder's summary alone can drift from the user need.
  • Paste the output, never describe it. A description is a claim about a result.
  • Preserve findings and responsibility. An independent validator reports defects to the owner before repair. An owner doing consumer checks fixes demonstrated in-scope causes and retains the failing evidence; no new refinement pass is needed.

Three questions, in order

  1. Does the demonstration do what the child claimed? Type the command or make the request. Read what it prints.
  2. Does it meet the acceptance criterion, or does it only run with no error? A criterion that passes without the behaviour is the failure to find.
  3. What does a consumer hit that a builder does not? Look for a missing prerequisite, an error that names an internal path, or an instruction that assumes this repository.

Use the operational environment

A consumer has no worktree, no fixtures and no knowledge of the implementation. When the change ships to a consumer, run it there: a new directory, an install, and the documented command. Some defects are visible only there, for example a file over a budget that the local config raised.

Rendered pages

A rendered page is an image. A page can pass every check and still be unreadable, so look at the picture:

  • Render the page and open the screenshot before you call it done. On WSL, drive Windows Chrome over a UNC path:
chrome.exe --headless=new --disable-gpu --window-size=1920,1080 --screenshot=<win-path> "file://\\\\wsl.localhost\\<distro>\\<path>"
  • Check each acceptance criterion against the picture.
  • Render from real data as well as the fixture. A fixture with fewer rows hides an overflow.
  • When the repository supplies an overflow checker, run it and read both of its signals, render-overflow and render-overlap. A clipped box shows no scrollbar.
uv run python .scripts/check_render_overflow.py <page.html>
  • Shoot every viewport width the layout claims.

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.