Dididecks-AI: DD-Ready Citation and Source-Access System

Stub — 2026-05-11. Sibling spec to [[Dididecks-AI-Slide-Decks-as-Code]]. Forked out preemptively per the anxiety-trigger principle in the context-vigilance skill: this is a substantial sub-system that deserves its own design space rather than bloating the parent spec.

The Question This Spec Will Answer

How does Dididecks-AI make every claim, chart, KPI, and key metric in a deck backable — with a citation that survives transmission, a source artifact the receiving party can independently retrieve, and a level of rigor that satisfies academic-grade scrutiny without breaking the authoring loop?

The North Star

A receiving-side decision-maker performing their own Due Diligence on a Dididecks-authored deck can click a citation, pull the original source, verify the claim, and continue their DD process without ever contacting the sender. That is the bar. Anything short of that fails the receiving-side DD test that the parent spec hangs its identity on.

Why this is its own spec (and not a section in the parent)

The parent spec, [[Dididecks-AI-Slide-Decks-as-Code]], commits to a Due Diligence Bar and lists “chart/KPI heavy with academic-grade citations” as Input #2 of the design rationale. But the citation-and-source-access system is large enough — in scope, in component primitives, in protocol design, in trust architecture — that putting it inline would bloat the parent past the comfortable-reading threshold. Forked out preemptively. The parent keeps the map; this spec carries the detail. See ~/.claude/skills/context-vigilance/references/philosophy.md → “On capability ≠ wisdom-to-use-it” for the rationale.

What this spec covers (placeholder section list — to be developed)

Headers only. Each will be filled through discussion.

The DD-Ready Citation Primitive

Chart, Table, and KPI Component Primitives

Infographic Layout Patterns

Source-Access Protocol (the receiver’s side)

Source Archival and Link-Rot Mitigation

Citation Capture Workflow

Receiving-Side UX

Prior Art

• lossless-monorepo/packages/lfm/ — the citation handling in @lossless-group/lfm, especially the hex-code footnote renumbering and tree.data.citations enrichment. Directly upstream.
• dididecks-ai/client-sites/calmstorm-decks/ — the most recent Dididecks-pattern site; its citation rendering is one half-step ahead of LFM’s general case. Worth studying for what extra it ships.
• astro-knots/sites/cilantro-site/ — strong reference for SEO/OG meta + content-collections architecture.
• astro-knots/sites/twf_site/ — cleanest LFM integration; parseContent polyfills for citation/callout handling.
• The chroma-agent-skills skill at lossless-monorepo/context-v/skills/chroma-agent-skills/ — semantic retrieval discipline that’s relevant when a deck draws from a large internal source corpus.
• The crawl-fetch-ingest skill at lossless-monorepo/context-v/skills/crawl-fetch-ingest/ — auto-populating citation metadata from a URL.

Open Questions

1. Citation schema — extend the existing LFM footnote shape, or design a new structured citation type that LFM adopts upstream?
2. PDF export fidelity — citations need to survive PDF export (per parent spec Design Principle #6). What rendering layer guarantees this? Native PDF generation (Playwright + print CSS, like calmstorm-decks does) or a separate citation-aware pipeline?
3. Archive timing — archive-on-cite (write-time) or archive-on-publish (publish-time)? Cost / fidelity / staleness tradeoffs.
4. Receiver-side authentication for paywalled sources — should Dididecks integrate with the receiving firm’s institutional access (Bloomberg, S&P Capital IQ, etc.) or stay neutral and route the user to their own login?
5. Bundled-skill packaging — does this whole sub-system ship as a single agent-skill (dididecks-citations) bundled into the workspace install, or as several composable skills (citation-capture, chart-with-source, source-archive)?

Related

• [[Dididecks-AI-Slide-Decks-as-Code]] — parent spec.
• [[Dididecks-AI-Visual-and-Diagram-Component-Library]] — adjacent sibling spec. Charts and infographics straddle both — citation-grounded quantitative content lives here; concept diagrams and mental models live there.
• [[../explorations/Dididecks-AI-Business-Model]] — relevant because citation/source-archival has hosted-cost implications.
• lossless-monorepo/packages/lfm/ — upstream code.
• lossless-flavored-markdown skill — the existing LFM conventions this builds on.
• context-vigilance skill — for the fork-and-cross-reference discipline this spec exists to honor.