agentleFS
devkindhqdevkindhq/ideogram-ai-toolkit/skills/seamless-pattern-texture-generator/SKILL.md

seamless-pattern-texture-generator

Skill3 starsChanged 12 days agoApache-2.0View on GitHub

Writes, generates, visually verifies, and iterates on seamless/tileable surface-design patterns — florals, geometrics, stripes, damask, plaids, novelty prints — and tileable material textures — fabric weaves, paper grain, wood, stone, leather, packaging substrates — for wallpaper, packaging, textile print, and digital surface design. Trigger on explicit requests using phrases like "seamless pattern," "tileable," "repeating pattern," "wallpaper design," "fabric print/texture," "packaging surface pattern," "surface design," or "this needs to tile edge to edge." Triggers on explicit request only — it does not auto-wire from `brand-identity-sheet`, `moodboard-generator`, or any other sibling skill; those skills are unmodified and unaware of this one, the same non-auto-wiring precedent `collections-management` sets for its own triggers.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 3 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Read all four reference files before running any workflow step:

    line 32
  2. Distinctive:

    If the user says "tile" without specifying straight, half-drop, or brick repeat, default to a straight repeat — and say so explicitly in …

    line 51
  3. Very specific:

    Follow references/tile-prompt-recipe.md.

    line 59
  4. Very specific:

    Before calling generate_image, score the drafted prompt against the pre-generation gate in references/anti-slop-discipline.md.

    line 68
  5. Very specific:

    Use mcp__ideogram__generate_image for a single tile.

    line 76
  6. Very specific:

    Write the JSON breakdown per the adapted references/composition-spec-format.md: motif instances and how they connect across the tile's …

    line 101
  7. Very specific:

    Write the prompt, the JSON, the verification result, and the image URL(s) to 04-projects/<project>/ — check for an existing …

    line 107
  8. Distinctive:

    Treat it as a failed generation requiring a prompt revision ("no border, no frame, no vignette, full-bleed edge to edge"), not something to …

    line 121

≈ “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

Seamless Pattern & Texture Generator

No generate_image, generate_images_bulk, remix_image, or reframe_image call in the connected Ideogram MCP exposes a seamless/tileable boolean, an edge-wrap mode, or an offset/repeat-preview capability — confirmed by direct inspection of those tools' schemas, not assumed from docs. Seamlessness here is achieved entirely through prompt language plus a best-effort visual-verification step (Workflow step 5). Never state or imply that a generated tile is mechanically guaranteed to be seamless — it is prompt-engineered and visually checked, not API-enforced.

style_type, negative_prompt, and seed are no-ops on the default v4 generation path this skill uses — they only take effect when custom_model_uri is set (the custom/v3 pipeline). Do not rely on them to set the DESIGN-vs-REALISTIC register or to exclude motifs/clichés in this skill's prompts; write the register and the exclusions directly into the prompt prose instead.

This skill covers two related deliverable families in one place — repeating surface-design patterns and tileable textures/materials — because both are the same underlying problem: one tile that has to read as continuous when it repeats. Structurally this is a prompt-composition skill like ideogram-prompt, logo-prompting, and moodboard-generator (one prompting problem plus a visual-verification loop), not a multi-call orchestration workflow like collections-management or custom-model-training.

Before you start

Read all four reference files before running any workflow step:

  • references/tile-prompt-recipe.md — the structured-caption recipe for writing the tile prompt itself.
  • references/tiling-verification.md — the visual-QA step used in Workflow step 5.
  • references/anti-slop-discipline.md — the pattern-specific slop bans and the pre-generation gate used in Workflow step 3.
  • references/composition-spec-format.md — the shared JSON schema, adapted here to motif instances, used in Workflow step 7.

Workflow

1. Gather the brief

Collect: repeat category (floral / geometric / stripe / damask / plaid / novelty / material-texture), end use (fabric / wallpaper / packaging / general surface digital use), colorway (exact hex if the user has one, a named palette if not), motif scale, and any reference image or material name the user is pointing at.

If the user says "tile" without specifying straight, half-drop, or brick repeat, default to a straight repeat — and say so explicitly in your response. State the reason: half-drop and brick repeats require offset math this skill has no crop/offset tool to verify, so silently defaulting to one without saying so would misrepresent what was actually produced.

2. Write the prompt

Follow references/tile-prompt-recipe.md. The prompt must state the seamless/tileable framing explicitly and early, include edge-continuity instructions, specify flat top-down framing with no directional lighting, exclude any border or vignette, and spell out the DESIGN-vs-REALISTIC register in prose — never via style_type, which is inert on the default v4 path this skill uses. If the user gave an exact hex palette, write the exact hex values into the prompt rather than a named-color approximation.

3. Run the anti-slop gate before generating

Before calling generate_image, score the drafted prompt against the pre-generation gate in references/anti-slop-discipline.md. Of everything on that gate, border/vignette presence and the copy-paste-rotation tell are the two checks worth failing the prompt over — both break tileability directly, not just aesthetics, so a prompt that fails either one gets revised before it's ever sent to the model.

4. Generate

Use mcp__ideogram__generate_image for a single tile. Use mcp__ideogram__generate_images_bulk (1-500 prompts) when the user wants several distinct colorways or scales rendered at once — batch them as one bulk call, not a loop of single generate_image calls. Default to aspect_ratio: "1x1" as the repeat-unit shape (a square tile) and rendering_speed: "QUALITY" for the deliverable render — edge-continuity is a composition/legibility-sensitive property the same way panel text is for moodboard-generator, and it's worth the slower render.

5. Verify

Per references/tiling-verification.md, generate a second, explicitly-labeled verification-only image asking for the same pattern shown repeated 2×2 with aligned edges, and report honestly whether seams are visible. State plainly that this is a visual approximation performed by the skill itself, not a mechanical guarantee, and do not imply the base tile is production-proven.

6. Style-lock variants

Once the user reacts to a base tile, use mcp__ideogram__remix_image for colorway/scale variations that keep the same motif. Use image_weight in the 60-80 range — tighter than moodboard-generator's 30-50 range — because a pattern variant needs to preserve the exact repeat structure, not just "same energy."

7. Compositional deconstruction

Write the JSON breakdown per the adapted references/composition-spec-format.md: motif instances and how they connect across the tile's edges, so the tile's structure is reusable later without re-describing it from scratch.

8. Save

Write the prompt, the JSON, the verification result, and the image URL(s) to 04-projects/<project>/ — check for an existing patterns/surface-design folder first and match whatever convention that project already uses. Nothing generated in this skill should live only in the conversation, per the "No Context Lost" rule.

Error handling

  • No native tiling flag exists anywhere in the connected surface. Never state or imply a generated tile is mechanically guaranteed seamless — always frame it as prompt-engineered and visually verified.
  • Verification image shows visible seams. Say so plainly and revise the prompt (stronger edge-continuity language, a simpler or larger motif, remove lighting/shadow cues) rather than shipping the seamed tile as if it passed.
  • Bordered or vignetted output (a common default reading of "pattern" prompts). Treat it as a failed generation requiring a prompt revision ("no border, no frame, no vignette, full-bleed edge to edge"), not something to upscale or ship as-is.
  • Ambiguous repeat type — the user says "tile" with no straight/half-drop/brick specified. Default to straight and state the default explicitly, per Workflow step 1.
  • style_type/negative_prompt set by mistake on a non-custom-model call. They are silently ignored by the API, not an error. Don't rely on them for register or exclusions on the default v4 path, and say so plainly if the user expects them to have an effect.
  • generate_images_bulk partial failures. Report per-prompt outcomes honestly rather than treating the whole batch as done if some prompts failed.

Reference files

  • references/tile-prompt-recipe.md — the structured-caption recipe for writing a seamless/tileable pattern or texture prompt.
  • references/tiling-verification.md — the 2×2 repeat visual-QA step and how to report the result honestly.
  • references/anti-slop-discipline.md — the pattern-specific slop bans and the pre-generation gate.
  • references/composition-spec-format.md — the JSON schema for the compositional deconstruction, adapted to motif instances.
  • examples/ — worked pattern/texture jobs go here as real jobs are run; empty for now.

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.