desi-webdesign-practicies-extraction
Desi Webdesign Practicies Extraction
Overview
Use this skill to turn a reference into a reusable reconstruction brief for another agent.
The input may be:
- a repo
- a component
- a code block
- a screenshot
- a live URL
The output is not a vague style description. It is a structured extraction of:
- what the design is doing visually
- how the screen is composed
- how motion and interaction behave
- which frontend mechanisms create those behaviors
- which exact source snippets should be carried forward when the logic is too complex to summarize loosely
Non-Negotiable Rule
Always store the original reference URL when one exists.
- If the user provides a live URL, store it verbatim.
- If the user provides a repo or component and a corresponding source URL is discoverable, store it.
- If no URL exists, record
No original reference URL provided.
Put this near the top of the output under Source Reference.
Workflow
- Identify the input type.
- Capture source metadata.
- Extract visible design language.
- Extract behavior and motion patterns.
- Extract frontend implementation patterns.
- Decide which mechanisms require code-backed teaching.
- Produce a reconstruction brief for another agent.
1. Identify The Input Type
Choose the extraction mode based on the strongest artifact available:
- Repo or full codebase: prioritize real implementation patterns and file-backed evidence.
- Component or code block: prioritize local logic, state, event handling, CSS structure, and rendering strategy.
- Screenshot or reference image: prioritize visual grammar and interaction hypotheses, but clearly mark inferred behaviors.
- Live URL: inspect rendered behavior first, then inspect code if available.
Prefer source-backed evidence over inference whenever code is available.
2. Capture Source Metadata
Start every extraction with:
Source Reference:original URL orNo original reference URL providedReference Type:repo, component, code block, screenshot, live URLArtifact Scope:full landing page, section, component, layout, interaction patternEvidence Level:source-backed, mixed, or inferred
When code is provided, also record:
- framework and stack
- key files
- component names
- relevant hooks or utilities
3. Extract Visible Design Language
Describe the page in precise screen-language terms, not generic adjectives.
Extract:
- atmosphere and vibe
- layout model
- layering system
- typography system
- surface treatment
- spacing rhythm
- color ownership
- information density
- asymmetry vs symmetry
Use concrete phrasing such as:
- full-screen horizontal chapter layout
- fixed atmospheric background
- glass instrumentation
- editorial typography over generative motion
- asymmetric list rhythm
- bottom-weighted hero composition
4. Extract Behavior And Motion
Explain what the user experiences:
- scrolling pattern
- navigation behavior
- section changes
- hover response
- cursor behavior
- reveal choreography
- loading sequence
- mobile gesture mapping
Focus on actual motion grammar:
- opacity plus translation
- stagger timing
- directional entry
- parallax or fixed layers
- scroll remapping
- pointer-follow physics
Avoid fuzzy phrases like "smooth animations" without saying what actually moves, when, and why.
5. Extract Frontend Implementation Patterns
For code-backed references, identify the mechanics that produce the design:
- state model
- refs
- event listeners
- scroll math
- gesture thresholds
- layout containers
- z-index and fixed-position strategy
- CSS variables and theme tokens
- reveal hooks
- animation loops
- pointer interpolation
- responsive layout branching
When useful, explain the relationship between structure and effect:
- fixed background plus moving content track
- hidden native cursor plus custom rendered cursor
IntersectionObserverplus transition classes for reveal-on-enter- pointer offset math plus
translate3dfor magnetic buttons
6. Use Code-Backed Teaching For Complex Patterns
Do not rely on natural-language summary alone when the behavior depends on implementation detail.
Include short source-backed snippets when the pattern involves:
- custom scroll remapping
- section index math
- shader or canvas readiness gating
- animation loops
- custom cursor interpolation
- magnetic button math
- gesture handling
- custom hooks
- stateful motion coordination
- layout calculations
For each snippet:
- keep it short and focused
- include file path or source location when available
- explain what the snippet is doing in plain language
- explain why the mechanism matters for recreating the effect
7. Mark Evidence Correctly
When the reference is image-only:
- separate what is visible from what is inferred
- label interaction and logic guesses as inferred
- avoid pretending you know the exact implementation
When the reference includes code:
- ground behavioral claims in the source
- cite files
- carry forward implementation patterns rather than only describing the result
Output Contract
Produce the extraction in this order:
Source ReferenceWhat This Reference IsDesign LanguageOn-Screen ElementsBehavior And MotionFrontend LogicComplex Patterns To Teach With CodeRecreation Instructions For Another AgentPrompt TemplateThings Not To Lose
References
Read references/extraction-checklist.md for the concrete extraction checklist.
Read references/output-template.md for the exact shape of the reconstruction brief.
More from michailbul/laniameda-skills
notion-sync
>
14instagram-extract
Extract text, links, and key takeaways from Instagram/Threads posts (especially carousels) and LinkedIn posts using an already-logged-in Brave/Chrome tab via OpenClaw Browser Relay. Use when the user pastes an Instagram/Threads/LinkedIn URL (or forwards screenshots) and asks something generic like “save it”, “capture this”, “summarize and store”, or “put this in the vault” — you should decide what’s worth saving, classify it into the right pillar, save it to the local KB and/or ingest prompts via the laniameda-kb skill, and also emit a compact JSON payload into a media-agent inbox file for downstream reuse.
12carousel-designer
>
11andromeda-messages
>
10deepgram-transcribe
>
10repo-kanban-pm
Install and enforce a lightweight product-management workflow inside a code repo: feature-as-kanban boards, ROADMAP status tracking, branch/PR conventions, and an optional daily OpenClaw cron PM review. Use this skill when (1) starting work in a NEW repository/project that does not yet have a ROADMAP+feature-KANBAN system, (2) you are delegated to spin up a new project and want multi-agent coordination from day one, or (3) you are asked to fix/restructure an existing repo and introduce an organized backlog/feature tracking system. Also use when adding a daily PM audit loop (cron) to keep code + docs + PRs in sync.
10