agentleFS
MyAiFreedomSystemsMyAiFreedomSystems/pre-flight-check/SKILL.md

pre-flight-check

Skill0 starsChanged 3 months agoMITView on GitHub

Mandatory pre-delivery quality gate for any artifact — websites, pages, images, documents, generated files. Use before delivering, shipping, deploying, or presenting any work as done, and whenever the user asks for a review, QA pass, or pre-launch check. Runs truth checks (no fabricated people, quotes, or numbers), copy review, versioning, link verification, geometry-based visual verification across the full breakpoint sweep (not just desktop and phone), a measurable review-team pass, and an honest delivery report. Prevents shipping unverified work.

3 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions3 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 3

    1. pskoett/pskoett-ai-skills★ 327same name
    2. mirekondro/The-Pre-Flight-Check★ 2same name
    3. Heverton-web/fabrica-universal★ 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.

The file

We split this skill into the rules it gives an agent

10 rules. 0 are common, 7 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 7

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Run every section in order before delivering anything.

    line 8
  2. Distinctive:

    Check deliverables AND their source files for text in a language the owner did not request (e.g.

    line 16
  3. Distinctive:

    Use direct-response, human phrasing throughout — headlines AND body copy, not headlines only.

    line 20
  4. Distinctive:

    Write product/skill names in Title Case everywhere.

    line 21
  5. Distinctive:

    Read the actual source material before summarizing it.

    line 22
  6. Very specific:

    Verify overlaps with geometry, not eyes: getBoundingClientRect() intersection tests between every floating/decorative element and every …

    line 27
  7. Very specific:

    Follow references/mobile-visual-audit.md for the true-phone-width harness, and references/responsive-breakpoint-audit.md for the sweep …

    line 29
  8. Distinctive:

    Verify every button and link resolves (curl or fetch each href).

    line 34
  9. Very specific:

    Never overwrite. Before changes: snapshot to versions/vN/ with a VERSION file.

    line 39
  10. Distinctive:

    Never present work as reviewed that was not.

    line 45

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “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

Pre-Flight Check

Run every section in order before delivering anything. A failing gate stops delivery — fix it or flag it, never ship past it.

1. Truth check

  • Trace every name, quote, number, and factual claim to a real source retrieved this session. If it does not trace, delete it or flag the gap to the owner — never fill gaps with invention.
  • Invented example material is allowed only when openly labeled as invented on every surface where it appears (visible caption, not just alt text).
  • Ban hype unless the owner wrote it: "free forever", "live now", license claims (e.g. MIT) that were not specified.
  • Public material contains no owner-internal names, machine details, or private infrastructure references.
  • Check deliverables AND their source files for text in a language the owner did not request (e.g. watermarks burned into generated images). Crop or regenerate before use.

2. Copy pass

  • Use direct-response, human phrasing throughout — headlines AND body copy, not headlines only.
  • Write product/skill names in Title Case everywhere.
  • Read the actual source material before summarizing it; state the real point, not a plausible one.

3. Visual verification (for anything rendered)

  • Enumerate the page's breakpoints first (grep -oE "max-width: *[0-9]+px|min-width: *[0-9]+px"), then screenshot at: desktop width (1440px), true phone width (390px), the owner's actual window width if known, one width inside every gap between breakpoints, and one pixel either side of each breakpoint value. A page that passes at 390 and 1440 can still be broken at 690 — mid-width bands are where absolute-positioned elements collide with stacked layouts.
  • Verify overlaps with geometry, not eyes: getBoundingClientRect() intersection tests between every floating/decorative element and every text, button, and card, at every tested width. Check scrollWidth <= innerWidth for horizontal overflow.
  • Frozen frames: occluded tabs freeze CSS mid-transition while computed styles show the final state. Take two screenshots 2–3 seconds apart before judging; never declare an element missing from a single capture.
  • Follow references/mobile-visual-audit.md for the true-phone-width harness, and references/responsive-breakpoint-audit.md for the sweep method, overlap snippet, and the CSS traps (mismatched breakpoints, transform-centering vs. animation).
  • Fix every cut-off, overflow, overlap, or invisible element, then re-shoot. Clean means clean at every tested width.

4. Links and claims

  • Verify every button and link resolves (curl or fetch each href).
  • Match counts and lists in copy against reality (platform counts, item counts — "twelve skills" must be twelve).

5. Versioning (for iterated projects)

  • Never overwrite. Before changes: snapshot to versions/vN/ with a VERSION file. After changes: add a CHANGELOG entry and bump any visible version footer.

6. Review team

  • When the owner has authorized subagent use, dispatch a reviewer subagent before delivering. Give it measurable acceptance criteria — exact viewport widths, the element lists that must not overlap, pixel tolerances, required evidence (geometry measurements + screenshot paths) — not "look it over." A review without numbers is an opinion, not a gate.
  • Fix every blocker it finds; fix nits or consciously waive them with a reason.
  • Never present work as reviewed that was not.

7. Delivery honesty

  • Build what was asked, literally. Do not add containers, frames, sections, or structure the owner did not request — if an instruction is ambiguous, ask before improvising structure around their content.
  • State in the final reply: what was done, what was verified (and how), what failed, what remains unverified. No confident language over unverified work.

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.