aparente gist - tufte-viz Claude Code skill
Why this is in the vault
Founder shared this gist via iMessage at 21:56 ET 2026-05-22 with no comment. Per feedback_sharp_verdicts_on_shared_content, treating as candidate reference. Filing because RDCO has multiple active data-viz surfaces (Sanity Check editorial charts, HQ generative-UI dashboards, MAC info-product visuals, HyperFrames composition art-direction) and zero codified viz-critique skill today. Direct gap-fill candidate.
What it is
A three-file Claude Code skill (SKILL.md + two references) for designing and critiquing data visualizations using Edward Tufte's principles. Covers data-ink ratio, lie factor, chartjunk elimination, small multiples, sparklines, layering, micro/macro design, and ships a 7-question Tufte Test plus a 14-question extended test for dashboard review.
Mapping against Ray Data Co
- Direct gap-fill: RDCO has
/design-criticfor general design surfaces and/video-criticfor HyperFrames renders, but nothing specifically for data-viz critique. A/tufte-vizanalog would fit the existing fresh-eyes-subagent pattern documented in [[feedback_fresh_eyes_subagent_for_own_artifacts]]. - Sanity Check editorial charts: Every issue that includes a data chart currently relies on ad-hoc taste. A Tufte checklist as a pre-publish gate (analogous to
/verify-vault-write) would catch lie-factor violations and chartjunk before the founder sees the proof. - HQ generative-UI substrate: hq.raydata.co dashboards built by the COO agent inherit no viz discipline. Codified Tufte principles in skill form = the agent applies them deterministically.
- Bookshelf gap reinforcement: [[2026-04-30-bookshelf-source-material-architecture-gap]] already calls out Tufte's Visual Display of Quantitative Information as a target source-material slot. This gist is the operationalized version of that book, ready to wire up.
- Install posture: Per [[feedback_mcp_install_security_review_default]], third-party skill install requires security review first. This is markdown-only (no code, no MCP, no hooks) so the review surface is small but should still run before adopting.
Related
- [[2026-04-30-bookshelf-source-material-architecture-gap]]
- [[feedback_fresh_eyes_subagent_for_own_artifacts]]
- [[feedback_mcp_install_security_review_default]]
- [[2026-04-15-commoncog-becoming-data-driven-first-principles]]
- [[../07-archive/readwise-thin/some-notes-on-executive-dashboards]]