agentleFS
chendanielychendaniely/skills/clothing-inventory/SKILL.md

clothing-inventory

Skill1 starsChanged 29 days agoMITView on GitHub

Use when adding or updating an owned garment in the `inventory/items/` register of a markdown knowledge vault — cataloguing a shirt, jacket, cardigan or similar from its photos and sewn-in labels, identifying it against the vendor''s own listing, and filing the label photos alongside the note. Not for pre-purchase research; that is product-search.

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    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:

    Write these sections, in this order.

    line 26
  2. Very specific:

    Start from template-clothing-inventory.md in this directory.

    line 131

≈ “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

Clothing Inventory Notes

Overview

The reader already owns it. They open the note to answer "what exactly is this, how do I wash it without ruining it, and where do I get another" — never "should I buy it". That last question belongs to product-search; keep the two apart.

Core principle: the garment is the primary source, the listing is corroboration. The sewn-in label is a legally required statement about this specific object; website copy is marketing about a product line, and it goes stale, gets carried over between seasons, and disagrees. When they conflict, the label wins and the conflict gets written down.

Second principle: empty beats estimated. A number, once written, is indistinguishable from a verified one. Uncertain facts go in prose where they can carry their caveat — never into frontmatter.

When to use

  • A garment is being added to inventory/items/ for the first time
  • An existing garment note is being corrected, or a label/measurement is finally read
  • Label photos have arrived for a garment already in the register

Not for: deciding what to buy (product-search), bags and non-clothing (the shared schema still applies — see inventory/Inventory.md — but the label/care/sizing machinery below is garment-specific), or anything in the vault's restricted areas.

The section contract

Write these sections, in this order. Omit one only if it is genuinely empty.

#SectionContains
1(frontmatter + H1)Full key block in the fixed order from inventory/Inventory.md, plus the garment deltas below
2Listing lineOne line, directly under the H1: official product link · price with currency · availability · date read. See below — this is non-negotiable
3Identification paragraphOne dense paragraph: construction and materials → brand, product name, product number → ownership and role → how and when identified
4Placement line"Nth garment in the inventory", naming the sibling notes and which of their precedents this one follows
5## What the labels sayOne bold sub-block per physical label. Spec label as a | Field | Value | table; care symbols as the 7-row table below
6## Identification: what is confirmed, and how| Fact | Source | Confidence | — values certain · high · unresolved · ⚠️ conflicted
7## What the vendor says about itBulleted, always stamped "read YYYY-MM-DD". Separate what the photos confirm from what only the copy claims
8## Measurements — size <X>Vendor size chart, all sizes, owned size bolded. Always carries the garment-not-body disclaimer
9(argued sections)Optional, one ## per real finding — a conflict, a care insight, why it earns its place. Name them for the conclusion, not the topic
10## NotesThe standing bullets — see Rules
11## Open questions- [ ] checkboxes, each naming its resolution method and which frontmatter field the answer fills
12## PhotosSplit ### The garment's own labels — <owner>'s photos, <date> from ### Reference shots — the vendor's, not this garment. Italic caption under every embed saying what it proves
13# ReferencesH1, not H2. External links with read dates, then internal wikilinks, ending [[Inventory]] — conventions

The listing line (non-negotiable)

Every garment note carries a clickable link to the official product page, directly under the H1 — not buried in References, where finding it costs a scroll and a scan.

**[Official listing](https://store.example.com/products/1026233011)** · ¥31,900 list / ¥29,800 sale · ⛔ sold out · read 2026-09-06
  1. Link to the brand's own page where possible; an authorised retailer second; an archive snapshot third
  2. Price with currency — the list price, never what was paid (that is cost, and it stays empty without a receipt)
  3. Availability, because a sold-out listing changes what "get another" means
  4. Date read — listings go stale and eventually 404

Mirror the URL into the url: frontmatter key so it is queryable, and keep the annotated entry in # References as the provenance record. All three, every time.

⚠️ When no official page exists, say so on that line and leave url: empty — **No official listing** — brand defunct; spec preserved below. A blank url: is a queue (the Needs product link view), a wrong one is a lie. Never guess a domain: a 200 proves a server answered, not that you reached the brand.

⭐ The listing will outlive its usefulness — copy the facts in, don't link and hope. Price, size chart, colourways, fabric and composition all get transcribed into the note. The link is provenance, not storage.

Frontmatter — deltas from the shared schema

inventory/Inventory.md owns the key list and their order. Do not restate it; do not contradict it. Garments differ from bags in exactly these ways:

KeyGarment rule
categoryThe vendor's own category word (shirt, cardigan, jacket) — even when it wears as something else. Say so in the note and cover the other reading in aliases
capacity_lOmitted entirely, not left blank. A bag field; a shirt has no volume. Blank means no published figure; absent means not applicable
sizeOff the size tab, cited. Empty until the tab is read (for example a piece confirmed from an outfit photo), with an ## Open questions checkbox to read it
aliasesGarment convention: the product number, the native-script and accented forms, and any wrong-but-intuitive name
colorThe observation, not the vendor's name for it. Vendor colourway names stay in prose until confirmed off the garment's own label
urlThe official product page. Empty if none exists — never a guess
costWhat was paid. A vendor list price is not a cost and never goes here

Note titles are ASCII (HERMES, COMME DES GARCONS); aliases carries the accented form (HERMÈS, COMME DES GARÇONS) so search finds either.

Care symbols — transcribe all of them, in label order

Japanese garments use JIS L0001 (ISO 3758-aligned). Build a \| \| Symbol \| Means \| table with one row per symbol.

⚠️ The bars are the part that gets missed, and they are instructions to a cleaner, not decoration — under Ⓕ or Ⓦ, one bar means mild, two means very mild. Count the iron's dots (1 = 110 °C, 2 = 150 °C, 3 = 200 °C) and check the drying square for a corner stripe (in the shade) as well as the bar (line dry).

⭐ Read the symbols as one argument, not a list. A run of fussy-looking instructions usually all defend the same property — loft, colour, or a stretch fibre's recovery. Say which, and name the one instruction that actually matters day to day.

Photos

The filename spec lives in inventory/Inventory.md. What is garment-specific:

  • Label photos are kind: serial, always with a -<detail> suffix — -brand-label, -spec-label, -care-label-front, -care-label-back. The care label is the garment's identifying document, the role a serial plate plays on a bag. Bare -serial-01 would be indistinguishable in a flattened export
  • Read every label at full resolution before transcribing. Crop and upscale the symbol block; a two-dot iron read as one dot is a 40 °C error
  • A vendor image is a stand-in, not a record. It clears the Needs photos view while proving nothing about condition, wear or even the right colourway. Flag it in the caption and in ## Notes, with the exact command to fix it: python3 scripts/file-inventory-photos.py "<Item Name>" --kind mine
  • A worn shot is filed as a face-free crop — the garment region only, the same as ootd files it
  • Verify the product image after downloading — read it back and check it shows this garment, in this colourway
  • ⭐ A white label in frame is a white-balance anchor. Black-versus-navy can often be settled from the photos already taken — sample the fabric, correct against the known-white label, and report the numbers. Do this before recommending anyone buy a lamp

Rules

  • Tables and bullets over prose. Prose only where reasoning genuinely needs sentences.
  • Trust the label over the website, and record the conflict rather than silently picking. Offer the competing explanations — a site error and a between-seasons fabric change are different stories with the same resolution.
  • Stamp every vendor claim with the date read. Prices, stock and size charts are perishable.
  • Unknown is a value. "Not published" and an empty field beat an estimate. Never write a plausible date, price or measurement.
  • Check the size chart for internal consistency before trusting it. If one column stops rising with size, it is a typo, not a cut — say which reading you believe and put "measure the garment" in ## Open questions.
  • Measurements are garment measurements, not body measurements — laid flat, chest doubled. Say it every time.
  • Supersede by annotation, never deletion. A corrected reading keeps the wrong one visible with the reason, because the wrong version is usually the intuitive one and will otherwise be re-derived.
  • Explain empty fields affirmatively. An empty evaluation: gets a Notes bullet saying the catalogs record pre-purchase research and this was not researched — otherwise it reads as an unfinished note.
  • Name the transferable finding. If something learned here would help next time, mark it ⭐ and state it as a rule in the note. Carrying it into the vault's CLAUDE.md or this skill is a proposal in the closing summary, applied only after the owner says yes. When run from ootd, ootd's write scope applies.
  • Score the garment against the vault's own frameworks where one applies, and resolve the tension honestly rather than scolding — a material that loses to the framework can still be right for the role.

Common mistakes

MistakeFix
Official URL only in # ReferencesListing line under the H1 + url: frontmatter + References. All three
url: filled with a guessed or retailer domainBrand's own page, or empty. A 200 is not proof of the brand
Price from the listing written into costcost is what was paid; list price lives in the listing line and the confidence table
Care symbols summarised as "hand wash"Transcribe all of them, in order, with the bars and dots counted
capacity_l: left blank on a garmentDelete the key — absent ≠ blank
Vendor photo with no caption disclaiming itSay "not a photo of the owned garment" in the caption and in ## Notes
Size chart copied without a sanity checkVerify each column rises with size; flag and measure if not
Colourway guessed from a photo's appearanceWhite-balance against a known-white reference in frame, or leave it open

Scope note — what generalises

Everything above the ## Care symbols heading applies to any owned item: the listing line, the url: key, the confidence table, empty-beats-estimated, supersede-by-annotation. The garment-specific parts are the care-symbol table, ## Measurements, the capacity_l omission and the label-as-serial photo convention. If a general inventory-item skill is ever factored out, that is the seam to cut along.

Template

Start from template-clothing-inventory.md in this directory.

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.