agentleFS
brunoasmbrunoasm/my_claude_skills/lab_ordering/SKILL.md

lab-ordering

Skill12 starsChanged 7 months agoApache-2.0View on GitHub

Place lab supply orders from member requests — route by request header to Amazon Business, the Pritzker Lab Google Form, or a direct vendor; stage the cart/form and stop for human review before any purchase. Use when the user pastes an order request or asks to order supplies, place an order, or fill the Pritzker form.

We split this skill into the rules it gives an agent

13 rules. About 0 are common, 11 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈11

    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. Distinctive:

    Use this skill when the user pastes a lab order request (a field table of Requested by / Item / Link / Part # / Quantity / Project-grant / …

    line 8
  2. Distinctive:

    Never complete a purchase or final form submission: Always stage the order and STOP for the user's review.

    line 17
  3. Distinctive:

    If the request is for the CLEAN LAB, emit a CAPITAL-LETTERS alert once the order is staged (see Clean Lab Alert below).

    line 22
  4. Very specific:

    If it exists, load identity, pritzker_form_url, amazon_business_url, and funding_sources.

    line 29
  5. Very specific:

    If it is missing, tell the user and walk them through creating it: copy the layout from references/config.example.yaml, ask for each value, …

    line 31
  6. Distinctive:

    Read the request header / first line:

    line 45
  7. Distinctive:

    Before touching a browser:

    line 55
  8. Distinctive:

    If the request's project/grant is "To be determined by Bruno", blank, or unclear, ask the user.

    line 57
  9. Distinctive:

    If a Claude-for-Chrome / computer-use browser tool is available, use the user's real (already-authenticated) Chrome.

    line 68
  10. Distinctive:

    If no browser control is available, fill in every value as text and give the user a ready-to-paste summary to drive the browser themselves.

    line 72
  11. Distinctive:

    Always begin the route by checking login state.

    line 75
  12. Distinctive:

    Follow the matching reference file.

    line 79
  13. Distinctive:

    If the request header/flags indicate the CLEAN LAB, after staging output a prominent capital-letters alert, e.g.:

    line 90

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 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

Lab Ordering Skill

Use this skill when the user pastes a lab order request (a field table of Requested by / Item / Link / Part # / Quantity / Project-grant / Urgency / Notes) or asks to place/stage a supply order.

Keywords: order, ordering, purchase, supplies, Amazon Business, Pritzker, cart, catalog number, restock, requisition.

Golden rules (never violate)

  1. Never complete a purchase or final form submission. Always stage the order and STOP for the user's review.
  2. Confirm the funding source with the user before staging any order.
  3. Flag every discrepancy (price vs. quote, out of stock, ambiguous item, login required, more items than a form holds).
  4. If the request is for the CLEAN LAB, emit a CAPITAL-LETTERS alert once the order is staged (see Clean Lab Alert below).

Step 0 — Load config (bootstrap if missing)

Read ~/.config/lab_ordering/config.yaml.

  • If it exists, load identity, pritzker_form_url, amazon_business_url, and funding_sources.
  • If it is missing, tell the user and walk them through creating it: copy the layout from references/config.example.yaml, ask for each value, and write the completed file to ~/.config/lab_ordering/config.yaml. Never store these values in the skill repo.

Step 1 — Parse the request

Extract, per item: item description, vendor/company, catalog/part #, quantity, product URL, plus request-level: requester name, urgency, notes/flags, and the stated project/grant. Field names are fuzzy — map synonyms (Part # = Catalog #, Project/grant = Account, etc.). A request may list multiple items.

Step 2 — Determine the route (decision tree)

Read the request header / first line:

  • Contains "Amazon" -> Amazon Business (see references/amazon_business.md)
  • Contains "Pritzker" -> Pritzker Google Form (see references/pritzker_form.md)
  • Otherwise -> Direct from vendor (see references/direct_vendor.md)

Note any flags in the header (e.g. "Clean Lab").

Step 3 — Confirm funding source

Before touching a browser:

  • If the request's project/grant is "To be determined by Bruno", blank, or unclear, ask the user.
  • Otherwise propose the best match from funding_sources and ask the user to confirm.
  • Carry the confirmed detail string forward:
    • Amazon -> PO field at checkout review.
    • Pritzker -> "Account to be charged".
    • Direct vendor -> PO/notes field if present, else the review summary.

Step 4 — Pick the browser method (environment-agnostic)

  1. If a Claude-for-Chrome / computer-use browser tool is available, use the user's real (already-authenticated) Chrome. Preferred.
  2. Else if in Claude Code, use the webapp-testing (Playwright) skill, and warn the user that logins will be required in the fresh browser.
  3. If no browser control is available, fill in every value as text and give the user a ready-to-paste summary to drive the browser themselves.

Always begin the route by checking login state; pause if not authenticated.

Step 5 — Execute the route

Follow the matching reference file. Default to the cheapest shipping unless the request notes ask for faster. Check availability/in-stock before staging.

Step 6 — Stop and report

Present a review summary table for each item: item, catalog #, quantity, price seen vs. quoted, link, funding account, shipping. List all flags. Then STOP and wait for the user — do not purchase or submit.

Clean Lab Alert

If the request header/flags indicate the CLEAN LAB, after staging output a prominent capital-letters alert, e.g.:

*** CLEAN LAB ORDER — THIS ORDER IS FOR THE CLEAN LAB. HANDLE / STORE ACCORDINGLY. ***

Available resources

  • references/pritzker_form.md — Pritzker form field mapping and fill steps.
  • references/amazon_business.md — Amazon Business cart + PO-field review.
  • references/direct_vendor.md — generic vendor cart-and-review.
  • references/config.example.yaml — placeholder config layout.

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.

Reviews can't be read right now.

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.