CtrlK
BlogDocsLog inGet started
Tessl Logo

brand-output

Branded output generation: HTML/CSS/JS, DOCX, XLSX, PDF, vector art, folios, templates (MetodologIA DS tokens in references/brand). Topics: brand-art, brand-docx, brand-html, brand-pdf, brand-xlsx, branded-html-output, folio-generator, html-brand, presentation-design, xlsx-template-creator.

SKILL.md
Quality
Evals
Security

brand-output

Router skill for branded deliverable generation with MetodologIA DS tokens. Resolve topic, then Read EXACTLY ONE playbook from routes: — never load the cluster. [EXPLICIT]

When to use

Any request to produce a branded artifact: HTML pages/folios, DOCX, XLSX, slides, or reusable templates. Not for raw content authoring (no brand) or non-MetodologIA output — route those elsewhere. [INFERENCE]

Inputs / Outputs

  • In: topic (one of the 8 enums), depth, the source content/data, target format.
  • Out: one branded file using references/brand tokens; provenance-tagged where claims aren't reproducible from the prompt. [CONFIG]

Routing

  1. Infer topic from the request; ask only if genuinely ambiguous (≥2 enums fit equally). [EXPLICIT]
  2. Read that single playbook. deep → apply exhaustively, verify each step; quick → essentials. [EXPLICIT]
  3. Format-vs-topic disambiguation:
    • HTML: brand-html (general) · html-brand (token-heavy/deep) · branded-html-output (lightweight) · folio-generator (paginated folios). [INFERENCE]
    • Slides → presentation-design. DOCX → brand-docx. XLSX → brand-xlsx (one-off) vs xlsx-template-creator (reusable template). [INFERENCE]

Spine: Discover → Analyze → Execute → Validate.

Router aids in assets/ — assets/routing-checklist.md and assets/quality-rubric.json (see assets/README.md). [CONFIG]

Validation gate (done = all true)

  • Exactly one playbook loaded; topic matches the artifact actually produced. [EXPLICIT]
  • MetodologIA DS tokens applied from references/brand; no hardcoded brand values. [CONFIG]
  • Constitution v6.0.0 enforcement + script-first rule honored (generate via script, not hand-edits). [EXPLICIT]
  • Evidence tags present on non-obvious claims; family not mixed (kit set: [EXPLICIT]/[INFERENCE]/[ASSUMPTION]/[CONFIG]/[DOC]). [DOC]

Anti-patterns

  • Loading multiple playbooks "to compare" — pick one, re-route only if wrong. [EXPLICIT]
  • Mixing brands in one output, or inventing tokens/prices. [EXPLICIT]
  • Skipping the script-first rule by hand-crafting the file. [ASSUMPTION]
Repository
JaviMontano/jm-adk-beta
Last updated
First committed

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.