Catalog an Item
Take new arrivals from shoebox to catalog: identify each item from its photo notes, record it with condition and an honest value basis, and flag what needs a better photo or an expert eye — uncertain identification stays uncertain in the record instead of becoming confident fiction. Locks the record fields and the honesty rule FIRST, then catalogs each item, then reports confidence and follow-ups. Use for cataloging new items, collection intake, 'catalog these arrivals', and adding finds to the collection.
Steps
Entry step: scope. Each step names the specialist role it wants; the full working prompt is expandable.
- Scope the intakecuratorentry
photo notes read, record fields and honesty rule locked
Show working prompt
Define the intake before recording anything. Step 1: read the photo intake notes waiting in inbox/ and the catalog store (collection.json) — the record fields in use and how existing items are shaped. Step 2: lock the record rules — every new item gets an id, name, category, condition, and its photo reference; a value estimate is welcome but only with its basis named; identification that is uncertain stays uncertain in the record ('pressing unverified'), never silently confident. Step 3: decide the follow-up buckets — needs a better photo, needs an expert eye, needs a document (receipt, certificate). Step 4: write an acceptance-criteria checklist ('every inbox item recorded exactly once', 'every record complete on the required fields', 'no identification stated above what the photos support', 'follow-ups named per item where earned'). Call write_task_note with the item count + rules + checklist and write notes/scope.md. - Catalog the arrivalscurator
identify each item from its photos, record it honestly
Show working prompt
Catalog each arrival against the locked rules. Step 1: for every photo note in inbox/, identify the item as far as the evidence carries — name, category, and condition from what the photos actually show. Step 2: record each item through the project's collection capabilities with id, name, category, condition, and photo reference; add a value estimate only where you can name its basis. Step 3: where a detail is unreadable or the identification is uncertain, record the uncertainty on the item itself and put it in the right follow-up bucket. Step 4: keep a per-item log in notes/catalog.md — what you identified, how confident, and why. Call write_task_note with the recorded count and the follow-up count.
- Write the intake reportcopywriter
items added, identification confidence, follow-up worklist
Show working prompt
Write the intake report from the catalog log. Step 1: open with the headline — items added this intake, out of how many that came in. Step 2: give a line per item — what it is, its condition, and your identification confidence in plain words. Step 3: list the follow-ups by bucket — better photo, expert eye, missing document — so the next session has a worklist. Step 4: flag anything whose value or identification deserves a second opinion before it is trusted. Keep it precise and unhurried — gallery-grade records are built one honest intake at a time. Write report.md and call write_task_note with the added count and the follow-up count.
- Evaluatereviewer
Grade the deliverable against every acceptance criterion. All pass → finish; any fail → loop back and fix the gap.
Show working prompt
Open report.md, notes/catalog.md, and the collection store and verify every criterion from notes/scope.md: every inbox item was recorded exactly once with the required fields complete, every identification stays within what the photo notes support (spot-check each against its inbox note), uncertainty is recorded on the item rather than smoothed over, and every earned follow-up is on the worklist. Write PASS/FAIL per criterion; on any failure, name the gap and loop back to catalog. Then route — this is the whole point of the loop: - **Every criterion PASSES →** call `advance_task_step({ ref, stepId: "evaluate", next: "finish" })`. - **Any criterion FAILS →** write the specific gaps to notes, then call `advance_task_step({ ref, stepId: "evaluate", next: "catalog" })` to loop back. The builder fixes exactly those gaps. Never route to `finish` while any criterion is unmet. The build phase's completion gate already blocked a grossly-incomplete deliverable; your job is the judgment an automated check cannot make (does it actually work, read well, look right). After ~3 unproductive loops, stop and report DONE_WITH_CONCERNS so the user can step in. - Finishdeveloper
All acceptance criteria met. Stamp a short summary and report DONE.
Show working prompt
Every acceptance criterion passed. Write a one-paragraph DONE summary to task notes via `write_task_note`: what was built, the deliverable path(s), and a one-line confirmation that each criterion is met. Then report DONE.
Triggers
Phrases that suggest this craftbook to a crew.
- catalog new items
- catalog these arrivals
- add items to the collection
- collection intake
- record the new finds
Source
View this craftbook on GitHub · MIT license