feat(creative): deck_builder v0.2.0 baseline — placeholders, quality linting, archetypes, and new layouts (#336)

[tusharjamunkar]

Summary

Part of #336 (v0.2.0 Baseline)

Implements the v0.2.0 baseline upgrade for creative/deck_builder, advancing it from a proof-of-concept slide assembler to an enterprise-grade presentation platform layer for autonomous agents as outlined in #336.

Following maintainer guidance (@rosspeili), this baseline PR establishes the core architectural foundations, layout extensions, heuristic linting gates, archetype outlines, and skill chaining pipelines in an incremental, reviewable step.

---

Scope & Roadmap Alignment

• Included in this Baseline PR:
  • Procedural neutral Pillow placeholders (hero, logo, icon, chart_backdrop, headshot) with zero remote dependencies.
  • Image fit policies (contain, cover, crop_center, stretch, native) with EXIF transposition and color-space normalization.
  • 3 new slide layouts: timeline, metrics, and comparison.
  • Presentation quality gates: lint_deck (scoring 0–100, 9 heuristic rules).
  • Deterministic archetype outline generator: suggest_outline (5 archetypes + constraint filtering).
  • Governance metadata ribbons (metadata.classification) and legal disclaimers (metadata.legal_footer).
  • Skill chaining demo (examples/deck_builder_chain_demo.py) and .skillware.yaml.example pipeline.
• Deferred to Follow-Up PRs (Phase 2b Roadmap):
  • extract_spec & render_patch actions.
  • New master templates, vibe tokens, and palette contrast validators.
  • team and agenda layout templates.
  • Per-layout strict required fields schema validation.
  • Golden PPTX corpus fixtures.

---

Key Capabilities Added

1. Procedural Neutral Image Placeholders & Fit Policies (placeholders.py):

   • 100% offline Pillow-based placeholder generator supporting hero, logo, icon, chart_backdrop, and headshot with subtle geometric motifs and typography badges.
   • Preserves offline-first constitution: zero remote network calls or image generation APIs inside the skill.
   • Pillow-backed EXIF transposition and color-space normalization (CMYK, Grayscale, RGBA to RGB).
   • Configurable image fit policies: contain, cover / crop_center, stretch, and native.

2. 3 New Slide Layout Types (timeline, metrics, comparison):

   • timeline: Horizontal milestone roadmaps with chronological connectors, dates, titles, descriptions, and color-coded status pills (completed, in_progress, planned).
   • metrics: KPI big-number cards (up to 4 per slide) with primary numeric callouts, metric labels, trend direction indicators (up, down, neutral), and delta badges.
   • comparison: Structured two-column comparative matrix supporting card dictionaries (title + items) or bullet arrays for contrasting features, tradeoffs, or before/after paradigms.

3. Presentation Quality Gates (lint_deck):

   • Deterministic heuristic deck scoring (0–100) evaluating presentation readability, bullet density, and layout rhythm.
   • Rule checks implemented: WALL_OF_TEXT, DECK_TOO_LONG, MISSING_ALT, ORPHAN_BULLET, EMPTY_TITLE, CHART_NO_TITLE, METRIC_WITHOUT_LABEL, LOW_SLIDE_COUNT, MONOTONOUS_LAYOUT.
   • Supports min_score and strict_a11y parameters declared in manifest.yaml.

4. Deterministic Archetype Outlines (suggest_outline):

   • Offline outline and skeleton generator for 5 standard presentation archetypes (investor_pitch, technical_brief, quarterly_review, product_launch, training_workshop).
   • Intelligent negative constraint filtering (e.g. constraints=["no pricing", "no financial"]) automatically removes sensitive sections while keeping narrative coherence.

5. Enterprise Governance Support:

   • Standardized deck_spec.metadata.classification header ribbons (CONFIDENTIAL, INTERNAL, PUBLIC, RESTRICTED) stamped on slides (with backward-compatible root fallback).
   • Standardized deck_spec.metadata.legal_footer disclaimer banner stamped on all slides.
   • Fully verified with explicit shape text assertions in test suite.

6. Skill Chaining Pipeline Demo & Tooling:

   • Authored examples/deck_builder_chain_demo.py showcasing composability with creative/bg_remover + creative/deck_builder (suggest_outline → bg_remover → lint_deck → render).
   • Added deck_build_pipeline named chain to .skillware.yaml.example.
   • Indexed demo in examples/README.md and added to tests/test_examples_smoke.py.

---

Verification & Test Coverage

• Bundle Unit Tests: 21/21 tests passed in skills/creative/deck_builder/test_skill.py.
• Integration Tests: 3/3 tests passed in tests/skills/creative/test_deck_builder.py.
• Total Skill Tests: 24/24 tests passed across unit and integration suites.
• Smoke Tests: 14/14 examples passed in tests/test_examples_smoke.py.
• Core Repositories: 92/92 tests passed across registry, docs, manifests, and UI schema suites.
• Linters & Formatters: black --check clean, flake8 clean (0 errors), scripts/sync_extras.py --check OK.
• Offline Invariant: Offline constitution validated (test_constitution_offline_no_remote_apis verifies no banned imports in any skill module).

[rosspeili]

Thanks @tusharjamunkar, very solid v0.2.0 baseline direction for #336. Architecture split (placeholders / lint / archetypes), offline constitution, chain demo, and targeted tests all look good on our side (23/23 deck_builder tests pass, and registry/docs/smoke clean).

Before merge:

1. Governance path bug, test_render_v020_new_layouts and instructions.md use root classification / legal_footer, but builder.py only reads deck_spec.metadata.*. Banners/footers are currently no-ops. Align contract (prefer metadata) and add a test that asserts ribbon + footer actually render.

2. brand_kit / theme.vibe, documented in schema/instructions but not implemented in builder.py. Either wire them up or remove from schema/instructions until phase 2b (don’t ship dead fields).

3. manifest.yaml, add parameters for lint_deck (min_score, strict_a11y) and suggest_outline (archetype, topic, constraints), plus outputs (score, passed, findings/issues, etc.).

4. Catalog ripples, update docs/skills/README.md to 0.2.0, append skill history row on deck_builder.md, and extend the chaining section to show suggest_outline → lint → render (you already have deck_builder_chain_demo.py).

5. PR description accuracy, bundle tests are 20 in test_skill.py (23 includes integration), please fix the count.

Scope note: This is a reasonable baseline (placeholders, fit, 3 layouts, lint, archetypes, chain). Full v0.2 (#336 parent roadmap) still defers extract_spec/render_patch, new templates/vibe/contrast, team/agenda, per-type required fields, golden fixtures, call that out in the PR so we don’t over-close the parent issue.

Happy to re-review after the additions. <3

[tusharjamunkar]

Thanks for the meticulous review @rosspeili! All 6 pre-merge items have been addressed in commit 07e7e78:

1. Governance path contract: Aligned \�uilder.py\ to read \deck_spec.metadata.classification\ and \deck_spec.metadata.legal_footer\ (with root fallback). Added \ est_governance_ribbon_and_footer_rendered\ inspecting shape texts directly to verify headers and footers render.
2. Dead fields removed: Removed \�rand_kit\ and \ heme.vibe\ from \deck_spec.schema.json\ and \instructions.md\ until Phase 2b.
3. \manifest.yaml\ parameters & outputs: Declared \min_score, \strict_a11y, \�rchetype, \ opic, \constraints\ parameters and all corresponding outputs.
4. Catalog ripples: Updated \docs/skills/README.md\ to 0.2.0, added 0.2.0 skill history row on \deck_builder.md, and extended the chaining section to demonstrate \suggest_outline\ → \�g_remover\ → \lint_deck\ →
   ender.
5. PR test count accuracy: Updated count in description (21 bundle unit tests in \ est_skill.py\ + 3 integration tests in \ est_deck_builder.py\ = 24 total).
6. Scope note & roadmap: Updated PR to \Part of [Skill Upgrade]: creative/deck_builder — v0.2 smart themes, images, and PPTX rework #336 (v0.2.0 Baseline)\ and explicitly documented deferred Phase 2b items (\extract_spec/
   ender_patch, new master templates/vibe, \ eam/\�genda\ layouts, golden fixtures).

All 24 deck_builder tests pass, and all 6 GitHub Actions CI checks are 100% green. Ready for your re-review! <3