AI Skill Report Card

Packaging Weaving Drafts into Digital Products

A-87·Oct 29, 2026·Source: Web
14 / 15

Given a WIF file or a scanned historical draft image, produce a sellable PDF pattern packet:

  1. Parse the draft (WIF preferred; vision-parse image as fallback).
  2. Validate it mechanically (shaft count, tie-up consistency, float lengths, hang-together check).
  3. Render a clean drawdown + threading/tie-up/treadling diagram.
  4. Add weaver-facing project info (sett, warp length, yarn estimate, floating selvedge note).
  5. Write historical/provenance context if the source is archival.
  6. Lay out as a print-ready PDF with correct DPI/bleed/margins.
  7. Write the listing copy and generate a mockup.

Output: a PDF the weaver can actually warp a loom from — not just a pretty picture of a grid.

Recommendation▾
Add a third example showing a fully successful, high-confidence case with complete output details (e.g., actual sett/yarn numbers) to balance the two edge-case examples given.
15 / 15

This skill sits downstream of pure draft-reading/analysis (that's a separate skill). Its job is productization: take a verified draft and turn it into something a customer pays for and can successfully weave.

Progress:
- [ ] 1. Source Intake — identify format (WIF, image, archival text) and extract raw data
- [ ] 2. Mechanical Validation — shaft count, tie-up type, float check, hang-together check
- [ ] 3. Provenance & Attribution Check — confirm public domain / licensing status
- [ ] 4. Drawdown + Diagram Rendering — clean, print-ready visuals
- [ ] 5. Weaver Specs — sett, warp length, yarn estimate, loom-type notes
- [ ] 6. Contextual Write-Up — historical note or design note (if applicable)
- [ ] 7. Layout & Format — PDF (print + screen), optional WIF re-export
- [ ] 8. Compliance Check — AI-disclosure, marketplace policy, no copyrighted motifs
- [ ] 9. Listing Copy & Visual Merchandising

Stage 1: Source Intake

  • WIF file → parse directly (threading, tie-up, treadling, warp/weft color, metadata fields).
  • Scanned image/screenshot → vision-parse the three sub-grids; flag low confidence rather than guessing.
  • Archival text description (e.g., 1800s weaving manuscript) → extract structural intent, then reconstruct a valid draft — do not assume a historical note is already loom-ready.

Stage 2: Mechanical Validation (non-skippable)

  • Recompute the full drawdown from threading × tie-up × treadling.
  • Confirm shaft count matches what's claimed in the title/description.
  • Run a float-length check appropriate to the structure (unit weave vs. non-unit block weave have different tolerances).
  • Run a hang-together check — confirm every warp and weft thread actually interlaces; a draft that "looks right" can still be unweavable.
  • If the source is a historical profile draft, confirm the correct block-substitution template was applied for the claimed structure (Summer & Winter, overshot, etc.) — do not re-publish an under-specified profile as if it were a complete draft.

Stage 3: Provenance & Attribution Check (non-skippable)

  • Confirm the source draft is public domain, licensed for reuse, or originated by the seller.
  • Historical pattern books (pre-1929 in the US, generally) are typically safe; named modern designers' drafts are not.
  • If reconstructing from a named historical source (e.g., a specific weaver's manuscript or a named archive), include attribution in the write-up — this adds sellable authenticity and avoids misrepresenting the work as original.

Stage 4: Drawdown + Diagram Rendering

  • Render threading (top), tie-up (small grid), treadling (right), and full drawdown as one clean composite image.
  • Produce both a black-and-white structural view (for weavers who want to see the true interlacement) and a colorway preview (for buyers evaluating the finished look).
  • Label shaft/treadle numbers explicitly — don't make the buyer count grid squares.

Stage 5: Weaver Specs

Include everything a weaver needs to actually warp the loom, not just admire the pattern:

  • Recommended sett (with structure-adjusted reasoning, not a blind WPI÷2 default)
  • Width in reed and finished width (accounting for weft contraction)
  • Warp length including take-up/shrinkage allowance
  • Yarn quantity estimate (warp + weft)
  • Floating selvedge requirement (yes/no, with placement note)
  • Minimum shaft/treadle count, and a table-loom lift-plan conversion as an optional add-on

Stage 6: Contextual Write-Up

  • For historical drafts: 2–4 sentences of verified provenance (era, region, structure name, original source). Do not invent historical claims — if provenance is uncertain, say "attributed to" or "in the style of," never fabricate a date or weaver's name.
  • For original/derivative drafts: a short design note (inspiration, intended use — towel, scarf, blanket).

Stage 7: Layout & Format

  • Print PDF: 300 DPI, correct trim size, 0.125" bleed, 0.25" safe margin, embedded fonts.
  • Screen PDF: smaller file size, clickable internal links (test every jump).
  • Optional: re-export a clean WIF file as a bonus asset — many serious weaver-customers want the loadable file in addition to the printed packet.

Stage 8: Compliance Check

  • AI-disclosure per marketplace policy if generative tools were used in layout/copy/mockups (not generally required for the draft math itself, which is deterministic).
  • No reproduction of a living designer's named pattern without license.
  • No trademarked guild/brand names in title or tags without rights.

Stage 9: Listing Copy & Visual Merchandising

  • Title: structure name + shaft count + project type (e.g., "8-Shaft Summer & Winter Table Runner Draft — Printable WIF + PDF").
  • Mockup: drawdown rendered as a photographed/woven-looking preview, plus a clean diagram thumbnail.
  • Tags/description pull from actual marketplace autocomplete, not generic weaving vocabulary.
Recommendation▾
Include a concrete template or checklist snippet for the weaver specs section (e.g., a sample table format) rather than just describing what to include.
15 / 20

Example 1: Input: A WIF file of a 4-shaft overshot coverlet pattern from an 1890s pattern book, no metadata. Output: Validated drawdown (float check passed), reconstructed sett/warp-length specs, a provenance note ("adapted from a 19th-century American coverlet overshot draft; original weaver unrecorded"), print-ready PDF + bonus WIF, listing titled "4-Shaft Overshot Coverlet Draft — Historical Pattern, Printable PDF + WIF File."

Example 2: Input: A blurry scanned photo of a profile draft labeled "Summer & Winter, 5 blocks" with no tie-up visible. Output: Flag as low-confidence — tie-up not visible, cannot verify shaft count or hang-together. Request a clearer scan or the WIF before proceeding to Stage 4. Do not publish a product from unverified source data.

Recommendation▾
Consider trimming some repetition between Workflow stage descriptions and Best Practices/Pitfalls sections, which restate similar points about validation and provenance.
  • Always re-derive the drawdown mathematically — never trust a stated structure name without verifying the actual interlacement.
  • Treat shaft-count and sett claims as testable, not descriptive flavor text.
  • Separate "what the draft is" (mechanical facts) from "what the draft means" (historical/design narrative) — verify the former, caveat the latter.
  • Give weavers the WIF file as a bonus whenever source format allows it; it's low-cost to include and meaningfully increases perceived value.
  • Default to under-claiming provenance ("attributed to," "in the style of") over fabricating specificity.
  • Publishing a profile draft as if it were a complete, loom-ready draft without applying block substitution.
  • Reusing a modern named designer's draft without checking licensing — historical public-domain status does not automatically extend to contemporary reinterpretations.
  • Shipping a drawdown image without the underlying specs (sett, warp length, yarn) — a picture of a grid is not a usable pattern.
  • Applying a single take-up/shrinkage percentage across all structures and fibers.
  • Skipping the hang-together/float check because the drawdown "looks fine" visually.
0
Grade A-AI Skill Framework
Scorecard
Criteria Breakdown
Quick Start
14/15
Workflow
15/15
Examples
15/20
Completeness
15/20
Format
15/15
Conciseness
13/15