nexu-io/open-design / plugins/_official/scenarios/od-next-strategy
od-next-strategy
VerifiedBundled OD Next V2 strategy entrypoint for request-led planning and Build execution.
assets/task-profiles/marketing.md
View on GitHub# OD Next Marketing Task Profile v2.0.4
> Rollout: active
Task routing, clarification, Build, and the ship-on-write boundary follow the
general orchestration Skill. Use the profile fields and delivery requirements as design guidance in the
readable plan; do not serialize them into a machine contract.
## Applicability
Apply this profile only to the parts of the current request that call for this
kind of deliverable. The Core instruction order applies: explicit user scope
and format outrank these defaults. An independent media or document request
does not require this profile's HTML source, page structure, or export workflow.
Keep the normal HTML requirements when building the corresponding prototype,
deck, marketing layout, or HyperFrames composition. Do not add a wrapper merely
because this profile is loaded.
## Profile fields
Resolve channel, dimensions and aspect ratio, audience, communication goal,
primary message and core selling point, call to action, brand assets, locked
copy, legal constraints, and required size variants. Freeze the visual
concept, palette, type, composition, image language, safe areas, and variant
rules in the Design Spec.
Never silently drop a missing field: decide per the general orchestration
Skill whether to enter the clarification stage, or convert the gap into
explicit assumptions with risks disclosed in prose. Legal and compliance
requirements must never be omitted, even when the user does not mention them.
Quality focus: first visual focal point, brand accuracy, channel fit, text
readability, clear CTA, and asset and factual fidelity.
## Artifact contract
For editable marketing layout work, the default deliverable is the source
for the chosen visual system: a single-file HTML source artifact with the
canvas sized to the channel spec, one canvas per size. Required deliverables list each final
channel asset and its derivation from the source. Multi-size work uses
composed variants, not an undifferentiated scale.
For this HTML layout route, product-side tools may render final images and
PDFs. This does not apply to direct image generation: an explicit standalone
image request uses the actual image-generation capability and delivers its
image file. For requested exports, use available tools or disclose a missing
capability; HTML alone does not fulfill an explicit image or PDF requirement.
For an HTML source request, writing the source is delivery; do not add
unrequested rendering or exports. Claim a final image or PDF only after that
file has actually been produced, regardless of which supported tool made it.
## Build Requirements
Global priority follows the Core System Prompt; this profile establishes no
separate order. Where upstream instructions and contracts are silent, carry
forward brand guidelines, the existing campaign system, and legal
requirements; user references and assets apply only within their designated
scope; this profile's defaults fill only what is left unspecified.
### Establish a communication focus recognizable at first glance
- Derive the primary message from the communication goal: brand, theme, core
selling point, campaign benefit, or CTA.
- Use scale, placement, contrast, and negative space to establish clear
first, second, and third reading levels.
- The main message must be recognizable at first glance — never depend on the
user reading all the fine print to understand it.
- Limit the number of selling points. When multiple messages sit side by
side, make the primary–secondary relationship explicit.
### Make visuals serve the copy
- The key visual must align semantically with the product, the theme, and the
copy; never add images or decoration unrelated to the message, included
only because they "look good."
- Match typeface style, graphic language, and color mood to the target
audience and the communication context.
- Never fabricate product capabilities, prices, discounts, awards,
endorsements, or campaign rules.
- **Licensed-asset boundary (scoped override):** for outward-facing marketing
collateral, this boundary overrides the core imagery-by-type rule for real
referents. Real photographs come only from user- or brand-supplied
licensed assets; a web-fetched image of a real entity is a disclosed
placeholder only, never final photography — and generating a fake stand-in
for a named real referent remains forbidden. Fictional or illustrative
subjects follow the core sourcing rule. When a required real
referent has no licensed asset, resolve the gap through the orchestration
missing-field policy — ask, or assume and disclose — rather than silently
using a fetched image as final or generating a fake.
### Keep brand and subjects intact
- Use the user-provided logo, brand colors, typefaces, product shots,
copyright information, and legal text correctly; never alter the logo's
proportions, colors, or required clear space.
- Keep products, people, and key subjects clearly visible — never broken by
text, decoration, or cropping.
- When no brand guidelines exist, establish a restrained, consistent
provisional visual language.
### Fit the publishing channel
When the task configuration specifies dimensions, the configuration wins;
otherwise use the channel's default spec, and state the spec used in the
delivery notes:
| Channel | Default spec | Key points |
|---|---|---|
| Xiaohongshu | 1080×1440 (3:4), optionally 1080×1080 | The cover image carries the entire click decision; title in the top 1/3; keep the bottom clear of the area covered by the like/comment bar |
| WeChat Official Account | header 900×383 (2.35:1), in-article images 900 wide | Header images get compressed in the feed; keep subject and text readable at thumbnail size |
| LinkedIn | single-image post 1200×627, article header 1200×644 | Information density may run higher than social norms; lean professional and data-forward; avoid decoration that feels overtly promotional |
| Offline poster | A3 297×420mm, or per venue requirements | Size type to viewing distance; a headline readable from 3 meters is no smaller than 70pt |
- Keep key information, the logo, faces, and the CTA within the central
70–80% of the canvas, ≥50px from the edges.
- Compose in three zones: brand or core value at the top, key visual and
supporting information in the middle, CTA at the bottom.
- When producing multiple sizes of one design, re-lay out the hierarchy and
composition for each; mechanical scaling is forbidden.
- Keep one creative core across channel versions while adapting to each
channel's viewing distance and information density.
### Guarantee readability at placement size
- Minimum type sizes on a 1080px canvas: headline 48px, subhead 32px, body
24px, captions 18px, CTA 28px; convert proportionally for other canvases.
- At most two typefaces per asset; body text vs background contrast ≥4.5:1.
- One CTA per asset: it starts with a verb, uses direct and specific
language, stands ≥44px tall, contrasts clearly with the background, and
sits where the eye path ends.
- For ad placement assets, cap the text share of the image — typically no
more than 20% of the frame.
- Choose type sizes during layout against the feed thumbnail and the actual
viewing distance, so the main message stays readable when scaled down to
50% — this holds at write time, never deferred to a later pass.
### Print delivery specs
Print parameters for offline collateral — 300 DPI (150 DPI acceptable for
large format), CMYK color mode, 3–5mm bleed on all sides, and pure black text
set in single-ink black rather than four-color overprint — are applied by
product-side rendering at export. The Agent builds the canvas at physical
dimensions in the HTML source artifact, keeps key content ≥5mm from the trim
line, and writes the print parameters above into the delivery notes and the
Build Package.
### Avoid templated AI visuals
- Apply the core anti-cliché defaults strictly: no warm beige / cream
default grounds and no generic stock-gradient backdrops unless the brand
or selected direction requires them.
- Avoid unmotivated neon gradients, glowing text, floating orbs, random
particles, and decorative 3D elements.
- Never invent awards, certification badges, press quotes, ratings, or
testimonials the user has not provided.
- Avoid over-rounded corners, excessive badges, and card stacks with no focal
point.
- Never use garbled or incorrect text, distorted logos, malformed hands,
misshapen product forms, or synthetic imagery that damages product or brand
fidelity.
- Design variants must differ genuinely in composition, message emphasis, or
visual concept — never by color swap alone.
- **Style-direction inference examples:** promotion pieces lean
high-saturation, high-density sales energy; brand-image pieces lean
photography-led generous negative space; event invitations lean symmetric
ceremony and refined finish. Without a brand, pick by scenario and vocalize
the choice in one sentence.
When the task configuration specifies a visual style, adopt it as the primary
visual approach and record it in the Design Spec: big-type poster styles make
typography the subject; photographic styles make the photo the subject with
text overlaid; gradient-block and geometric-abstract styles carry the
information through graphic structure. The chosen style neither exempts the
anti-cliché rules above nor overrides brand guidelines.
### Quality floor
Meet the following in one pass, while writing the source:
- Font and asset references are real and usable; the layout produces no
overflow or clipping; subjects are never covered by text.
- Canvas dimensions map one-to-one to channel specs; key elements sit inside
the safe area.
- Copy, prices, dates, and campaign rules are taken verbatim from
user-provided content; logo proportions and clear space stay as given.
- Each multi-size version re-lays out hierarchy and composition individually;
no mechanical scaling.
- Print collateral builds its canvas at physical dimensions with bleed, and
the print parameters go into the delivery notes; never describe a final
image or PDF that Open Design's engineering has not yet rendered as
completed.
## Build Packages
Use simple mode for one concept and a small coherent variant set. Complex
mode may split independent creative directions or dimension groups only after
the key visual, core message, brand rules, source assets, and Design Spec are
frozen; mechanical scaling is forbidden. Every package must name its exact
output sizes and shared source dependency.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| planning-and-task-breakdownaddyosmani/agent-skillsBreaks work into ordered tasks. Use when you have a spec or clear requirements and need to break work into implementable tasks. Use when a task feels too large to start, when you need to estimate scope, or when parallel work is possible. | Stars: 119k | Updated | Claude CodeCodex | Verified | Install |
| screenpipethedotmack/claude-memSearch screen and audio history recorded by Screenpipe and optionally save selected context into claude-mem. Use when the user asks to recall desktop activity, a meeting, or something seen in another app through Screenpipe. | Stars: 118k | Updated | Claude CodeCodex | Verified | Install |
| contact-widgetnexu-io/open-designSelf-contained floating chat widget with welcome screen, social links, meeting button, and message input. Single HTML file, zero dependencies. | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |
| meeting-notesnexu-io/open-designMeeting notes page — title bar with attendees, agenda checklist, decisions block, action items table with owners + dates, and a "next meeting" footer. Use when the brief mentions "meeting notes", "minutes", "1:1 notes", "all-hands recap", or "会议纪要". | Stars: 108k | Updated | Claude CodeCodex | Verified | Install |