# Minimal Editorial Exports > An alternate visual style for aria-code's own generated exports — report covers, PPTX title slides, single-chart share images, Canva one-pagers — built on heavy negative space, one accent color, and restrained serif/monospace type, instead of the default dense institutional/dashboard look. Trigger for "极简", "杂志风", "editorial", "zine", "less corporate", "make the cover page quieter", "one number, lots of white space", or when the user wants a report/deck/chart export that reads as considered rather than templated-dashboard. Do NOT trigger for the default financial report body (tables, indicators, full analysis) — this governs cover pages, title slides, and single-focal-point exports only, never dense data views (use `terminal-software-design` or the existing institutional/bloomberg themes for those). Do NOT trigger to invent a generic aesthetic from nothing — use `ui-ux-pro-max` for open-ended exploration instead. - Skill: `artherahq/minimal-editorial-exports` (Agent Skill, multi-file: 5 files) - Install (CLI): `npx skillmds@latest add artherahq/minimal-editorial-exports` - Raw SKILL.md: https://api.skillmd.com/api/skills/artherahq/minimal-editorial-exports/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Docs & Writing - Author: artherahq (https://skillmd.com/u/artherahq) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/artherahq/minimal-editorial-exports --- # Minimal Editorial Exports Financial report tooling defaults to density — dark tables, corporate blue, every indicator visible at once — because that's what the *body* of a report needs. But aria-code also produces things that aren't report bodies: a cover page, a PPTX title slide, a single chart meant to stand alone, a Canva one-pager. Applying the dense-dashboard default to those is a mismatch: a cover page crammed with the same density as the analysis behind it reads as unfinished, not thorough. The fix is a second, deliberately restrained style for exactly these single-focal-point surfaces — heavy negative space, one accent color, one number or one chart doing the work, quiet typography carrying the rest. Silence read as confidence in print design long before software existed; the same logic applies here. ## When this matters - A report's cover page or first slide, before the dense analysis begins - A PPTX title slide (`aria.report.pptx` / `report_exporters.py`) - A single chart meant to be shared standalone, not embedded in a full report - A Canva brand-template autofill for a one-pager or social export - Never the report *body* — tables, multi-indicator panels, and the institutional/bloomberg themes stay dense on purpose; this style is for the one surface per document that's allowed to be quiet ## Design rules 1. **70%+ quiet space.** The cover/title surface should be mostly negative space around one focal point — one number (e.g. the headline return), one chart, or one signal — not a summary of everything the document contains. 2. **One accent color, tied to real meaning.** If there's a directional signal, the accent is that signal's real color (gain/loss, buy/sell) — never a decorative color chosen for mood. If there's no signal, pick one restrained anchor color and don't reach for a second. 3. **Restrained type, not the safe default.** Serif or monospace over the generic sans-serif dashboard default; one size doing the heavy lifting (the headline number or title), everything else quiet caption-weight. 4. **Texture only where the format supports it.** An HTML/PDF cover (`report_generator.py`'s `export_pdf` path) can carry a subtle paper-grain or noise background; PPTX/DOCX covers (`report_exporters.py`) can't render texture the same way — for those, let space and type do the work instead of trying to fake grain. 5. **No borrowed dashboard chrome on the cover.** No sidebar, no nav, no multi-panel grid — a cover that reuses the report body's chrome isn't restrained, it's just a smaller dashboard. ## Applying this in aria-code - **`report_generator.py`** — `generate_price_chart`'s `mpf.make_marketcolors(...)` call sets the chart palette; for a standalone/cover chart, drop it to a single accent + neutral rather than the full up/down + volume palette used in the report body. - **`pdf_report.py`** — `THEMES` (`institutional`, `bloomberg`) are both dense-body themes; this style is for the *cover page* those themes render before the body starts, not a competing third dense theme. - **`report_exporters.py`** — `markdown_to_pptx`'s title slide (the first slide, before per-heading slides start) is the one slide this style governs; keep it to one line of restrained type and real negative space, not a bullet list. Same for `markdown_to_docx`'s cover heading. - **`canva_client.py`** — when building the brand template this style targets, keep the template itself minimal (one text field, one optional image/chart field); `autofill_design`'s `data` argument should carry one headline value, not a dense field set. ## Variation axes — don't let every export look identical Run this style enough times (one cover per symbol, one title slide per report) and a single fixed recipe becomes its own template — the exact genericness problem `ai-generated-ui-craft` warns about, just at the style level instead of the whole-page level. The fix isn't random variation (a randomly-placed layout on a financial document reads as arbitrary, not considered) — it's varying along a few independent axes, each driven by something real about *this* export, not a dice roll: - **Focal-point position** — centered / lower-third / upper-block. Derive from something real when possible (e.g. a strong directional signal implies weight toward one side) rather than defaulting to centered every time purely out of habit. - **Focal-point type** — one number (headline return/price), one chart (`aria.report.chart`'s output), or one short line of text — pick whichever actually carries this specific report's story; don't reach for "always the chart" or "always the number" regardless of what the report is about. - **Type pairing** — serif-only, monospace-only, or a serif headline + monospace caption pair. Rotate deliberately across a batch (e.g. a multi-symbol daily digest) rather than letting every export in the same run pick the same pairing by default. - **Quiet ratio** — 70% vs 85%+ negative space is itself a choice, not just a floor; a headline-only cover can afford more silence than one carrying a chart. None of these axes touch the accent-color rule above — the accent stays tied to real signal meaning regardless of which layout/type/quiet-ratio combination gets picked. ## Automated Quality Gate `scripts/exports_gate.py` turns the checklist below into code, the same pattern `minimal-editorial-poster`'s `poster_gate.py` uses (these two skills are siblings — keep the two gates' shape aligned if one changes). Build a spec with `surface`, `export_format`, `quiet_pct`, `focal_point`, `focal_point_type`, `accent_color` (`{name, hex, source}`), `typography`, `texture`, `chrome_notes`, and — when part of a batch — `is_batch`/`variation_axis`/`variation_reason`, then: ```bash python scripts/exports_gate.py --spec spec.json python scripts/exports_gate.py --demo ``` `--demo` compiles a dense cover wearing a "minimal" label (25% quiet space, borrowed sidebar/nav chrome, texture on a PPTX slide that can't render it, Inter/system-ui type) next to the corrected minimal-editorial compile for the same report — the first FAILs with named codes (`insufficient_quiet_space`, `borrowed_dashboard_chrome`, `texture_unsupported_by_format`, …), the second PASSes. ## Checklist - [ ] Is at least 70% of the cover/title surface genuinely quiet, not just slightly less dense than the body? - [ ] Is the one accent color tied to something real (a signal), or at least chosen once and not fought by a second color? - [ ] Is the type restrained and specific, not the default dashboard sans? - [ ] Did texture only get used where the export format can actually render it (HTML/PDF), not faked on PPTX/DOCX? - [ ] Does the cover avoid reusing the report body's chrome (sidebar, nav, multi-panel grid)? - [ ] Is this being applied to a cover/title/single-focal surface, not the dense report body itself? - [ ] If generating more than one export in the same run/batch, does the focal-point position, focal-point type, or type pairing actually vary across them — driven by something real, not left on autopilot?