Pages
Report Cards — Mini-doc Ideas

Source: c78 fleet-overview-reports — mini-document expansion (iteration on report-card-ideas §4), 2026-08-24

Stage 4's mini-document had the strongest identity and the weakest instrument. This iteration keeps the white page and makes typographic hierarchy the instrument: of everything a report knows — score, worst item, movement, verdict, location, identity, recency — which single fact deserves the biggest type? Seven hypotheses, each shown as its full four-card row. Every value is derived (deriveCardSummaries() + fixture structure); the paper stays white in both themes (artifact identity), so on-paper ink is fixed, never themed. One attention device per card. Judging law unchanged: "do I need to open this report?"

1

Headline — the score is the front page

Hypothesis: state outranks everything. The score set like a front-page numeral — largest type on the page by 4× — title demoted to a kicker, delta as the subhead. The stage-4 register treatment, wearing paper.

Compliance
83%
▼4 vs prior month
CMP-2026-08
SLA Insights
89%
▼5 vs prior month
SLA-2026-08
Inventory
87%
▼2 vs prior month
INV-2026-08
Board Report
86
▼4 vs prior month
BRD-2026-Q2

Notes: the strongest four-up scan of the seven — four numerals read like a dashboard even at thumbnail size. But it inherits the register's weakness: an 89 and an 83 read as neighbors when one hides a failing control; tone color is the only urgency channel.

2

Lede — the worst item is the opening sentence

Hypothesis: the sentence outranks the number. Memo typography — title as heading, the report's most urgent line set as a real, readable bold lede; the rest of the document greeked. A clean world reads a calm lede. Score demoted to the running header.

Compliance83%

CC6.1 failing — Open high finding — secrets.yaml read in an agent session (maps to LLM06)

CMP-2026-08
SLA Insights89%

4 open, all within SLA

SLA-2026-08
Inventory87%

5 shadow installs — newest Windsurf on chen-mba

INV-2026-08
Board Report86

AI governance posture: 83% of controls compliant across 2 frameworks · 1 open high finding · 87% of the agent estate monitored

BRD-2026-Q2

Notes: answers the judging question most directly — the card tells you why to open the report, in words. Costs: 9px is the floor of honest legibility, and the board card's "worst item" is its whole headline, which needs a hard clamp to stay on the page.

3

Stamp — the verdict as an audit mark

Hypothesis: one bit outranks all detail. The document itself stays fully greeked and quiet; the verdict lands on top as a rotated, letterspaced audit stamp. Weight and tone carry the whole message — the paperwork idiom for "reviewed."

Compliance
CMP-2026-08
2 failing
SLA Insights
SLA-2026-08
Current
Inventory
INV-2026-08
5 shadow
Board Report
BRD-2026-Q2
Attention

Notes: the loudest peripheral signal of the seven — an amber or red stamp is legible before the card is even read, and a quiet green CURRENT is the calmest all-clear. But it's a single bit: no score, no movement, no reason. The stamp says whether, never why.

4

Contents — the trouble located inside the document

Hypothesis: where outranks what. A miniature table of contents, derived from the report's real structure — each section with its figure. The one section that needs you is bold and toned; healthy sections stay faint. The card answers "do I open it" and "where do I land."

Compliance83%
SOC 2 Type II5/6 compliant
NIST 800-53 (moderate)5/6 compliant
Evidence1279 items
Snapshots6 retained
CMP-2026-08
SLA Insights89%
Targets4 severities
Quarters4 rollups
Findings4 open · 0 past target
Snapshots4 retained
SLA-2026-08
Inventory87%
Harnesses39 installs
Models6 in use
MCP servers9 connected
Shadow5 unmonitored
INV-2026-08
Board Report86
Controls compliant83%
Open past SLA0
Estate monitored87%
Snapshots2 packets
BRD-2026-Q2

Notes: the most information per square pixel, and the only candidate that names the section you'd drill into — the TOC row is a natural deep link. Busiest of the seven; at 8px the rows are labels more than reading text, and four TOCs side by side no longer scan as a dashboard.

5

Marginalia — the page opened to the finding

Hypothesis: context outranks summary. Not the cover — an inner page, mid-document: running header, greeked paragraphs, and one real line highlighted where the worst item sits. The card shows the exact line you'd land on. A clean world highlights nothing.

Compliancep. 4

CC6.1 failing — Open high finding — secrets.yaml read in an agent session (maps to LLM06)

CMP-2026-0883%
SLA Insightsp. 3

Nothing flagged this period.

SLA-2026-0889%
Inventoryp. 5

5 shadow installs — newest Windsurf on chen-mba

INV-2026-0887%
Board Reportp. 2

AI governance posture: 83% of controls compliant across 2 frameworks · 1 open high finding · 87% of the agent estate monitored

BRD-2026-Q286

Notes: the most document of the seven — it looks like a report someone has already read for you, highlighter in hand. Weakest at four-up scan: all four pages look alike until you read the highlight, and identity drops to a 7px running header.

6

Cover — identity first, formalized

Hypothesis: the file is the message. The stage-4 original taken to its formal conclusion — a report cover: centered title as the largest type, generous whitespace, score set beneath, run id as the filing line. The control condition the other six are judged against.

Auditors
Compliance
83%
observation period · 6 mo
CMP-2026-08 · Aug 1
Compliance & security teams
SLA Insights
89%
observation period · 6 mo
SLA-2026-08 · Aug 1
IT ops & security teams
Inventory
87%
observation period · 6 mo
INV-2026-08 · Aug 1
Executives & board
Board Report
86
observation period · 6 mo
BRD-2026-Q2 · Jul 5

Notes: the purest "your landing is four files" statement, and the calmest row on this page. As an instrument it's still the stage-4 verdict: identity over density — state lives in one toned numeral, urgency and movement are absent.

7

Ticker — movement over state

Hypothesis: the needle outranks the number — Mark M's "we wanna know if we're moving the needle," typeset. The delta is the largest type; the current score rides beside it; the trend is embedded in the page as a document figure.

Compliance
▼4now 83%

6 mo
CMP-2026-08
SLA Insights
▼5now 89%

6 mo
SLA-2026-08
Inventory
▼2now 87%

6 mo
INV-2026-08
Board Report
▼4now 86

6 mo
BRD-2026-Q2

Notes: the only candidate built around the cycle's time-series consideration — a fleet getting worse shows it here first. But direction isn't urgency: compliance fell 4 points and has a failing control, while a card that fell from 97 to 94 looks identical in kind. The delta's tone answers "which way," not "how bad."

Provisional verdict — awaiting Mark's pick

  • The hierarchy is the instrument. Stage 4's mini-doc failed because state hid in a caption; every variant here fixes that by promoting one derived fact to the top of the type scale. The disagreement is only over which fact.
  • Four-up scan favors Headline; the judging law favors Lede. Headline reads as a dashboard at thumbnail size; Lede is the only form that says why to open the report. A Headline spine with the Lede sentence beneath it (score large, sentence small-bold) is the natural merge — the paper edition of stage 4's register + worst-line verdict.
  • Stamp is the best alert-world dressing, not a card. Layering it over any other variant doubles the attention devices; if it's used, it must replace the tone color, not join it.
  • Contents earns a different home. Location-grade detail belongs where drill happens — it's a hover state or the report's own nav, not landing altitude. Marginalia likewise: strongest as the artifact preview on export/shelf surfaces.
  • Ticker honors the time-series consideration but confuses direction with urgency; its embedded-figure sparkline is worth stealing into whichever form wins.
  • Final pick is Mark's; whichever wins should be re-composed into the fleet-overview-v3 spine for a side-by-side with the current merged treatment.
  • Iterated: the greek lines are retired in report-card-minidocs-v2 — every slot on the card is real, audited content.

Qpoint Brand Style Guide