← Vault Index
Source: frameworks/kit-field-guide-production/04-field-guide-quality.md

Quality — Field Guide QC Checklists

Gate 1: Pre-build — all editorial decisions confirmed before building starts.

Gate 2: Post-build — 100 points, pass threshold 90/100. Run after every build and after every revision.

Both gates must pass before any field guide is delivered.


Gate 1 — Pre-Build (Editorial Decisions)

Run this before opening the build skill. Any open item stops the build.

Routing Check — Run This First

Has the consultant completed the book analysis (File 07)?

Editorial Decisions


Gate 2 — Post-Build (100 points, 90+ to pass)

Run in this order.


Prompt Quality (30 points)

#CheckPoints
1Every prompt tested by pasting into Claude — AI introduces itself, lists steps, walks through the full guided session, produces the named deliverable10
2Every prompt follows TASK/INTRODUCTION/STEP format with 5 steps4
3Every step ends with a confirmation question that pushes for honesty4
4Session handoffs are graceful — work whether or not the reader completed the previous session4
5Final prompt constrains output — "one thing per dimension" or equivalent3
6Every instruction line ends with "The AI will guide you through each step"2
7No prompt references a session that hasn't been completed yet3

Prompt testing is blocking. A field guide with untested prompts does not ship.


Interactive Functionality (20 points)

#CheckPoints
8Every interactive tested through 3+ scenarios with different input patterns6
9Summary/result text appears at correct threshold3
10Math is correct — percentages, averages, counts all verify manually4
11Edge cases handled — all selected, none selected, tied scores3
12Verdict/summary text is accurate for each scenario4

Interactive testing is blocking. Every scenario must be clicked through in a browser.


Copy QC (10 points)

#CheckPoints
13No twinning — "You don't have X. You have Y." (P1)2
14No mirror reversals (P1)2
15No three-beat parallel lists (P2)1
16No "Not because X. Because Y." (P2)1
17No rhetorical hand-holding — "Right?" "Sound familiar?" (P3)1
18No correction-revelation compound in same slide (P1)3

P1 copy failures are blocking.


Brand Compliance (20 points)

#CheckPoints
19Zero rgba() in the entire file — run grep -c 'rgba(' file.html and confirm 05
20Zero CSS variables — grep -c 'var(--' file.html confirms 02
21Gold text on light backgrounds uses #6b5d3e — never #b79d643
22No pure black (#000000) — always #1a1a1a or darker named hex1
23Typography uses rem and clamp() — not fixed px for content text2
24Cormorant Garamond font URL loads ONLY weights 300, 400, 500 — no 600/7002
25Eyebrow weight is 600. Strong text weight is 500.1
26Copy button copied state: dark bg (#1a1a1a) + gold text (#b79d64)1
27Gold accent bar present on slide titles (::after, 40px × 3px, #b79d64)2
28Scrollbar uses solid hex (#d4c9a8 thumb, #f5f4f0 track)1

Zero rgba() is blocking. A file with any rgba() values does not ship regardless of total score.


Structure, Layout, and Visual (20 points)

#CheckPoints
2914 slides in correct order — Intro, Diagnostic, Mental Model, 5 idea pairs, Bridge3
30Mental model slide (3) has two-column comparison + dark insight callout2
31Attribution line reads "Based on the book by [Author Name]"2
32Author name does NOT appear in header, title treatment, or campaign label1
33Diagnostic result says "will hit hardest" not "start with"1
34All slides use full-width stacked layout — text in container, interactive in container-wide3
35Exercise instructions merged into frame header, not floating as separate callout2
36Bridge has single centered CTA on dark background — no offer cards2
37Visual rhythm: session slides on off-white (#faf9f7), bridge on dark (#1a1a1a)2
38Rating exercises use stoplight colors (red #c45a4a → orange #d48a3c → yellow #e0c34a → green #5a9a6a)1
39Slider exercises have anchor labels describing what each end means1

Visual verification sub-checks (must complete to claim points for 29 and 34):

Visual verification is blocking. A file that has not been opened and clicked through does not ship.


Scoring Summary

CategoryPoints
Prompt Quality30
Interactive Functionality20
Copy QC10
Brand Compliance20
Structure, Layout, and Visual20
Total100

Pass threshold: 90/100

Blocking failures (must fix regardless of score):

After any revision: Return to Gate 2 and run all checks again.


Common Failure Modes

FailureWhat It Looks LikeFix
rgba() remaininggrep -c 'rgba(' returns anything > 0Replace every instance using hex table in File 03
Two-column layoutInteractive crammed into right columnText in container, interactive in container-wide below
Exercise instruction disconnectedFrame header says "Your Exercise," instruction floats aboveMerge: "YOUR EXERCISE — [what to do]"
Gold on cream text#b79d64 text invisible on creamAll gold text on light = #6b5d3e
Bracket-field promptsOld fill-in-the-blanks formatTASK/INTRODUCTION/STEP with 5 steps
Slider without anchorsReader doesn't know what 1 or 10 meansAnchor labels under every slider
Rating without stoplightColors don't communicate severityRed → orange → yellow → green
Multiple bridge offersDeploy Sprint / Advisory OS cardsSingle centered CTA on dark background
No visual rhythmAll slides same cream backgroundSession slides off-white, bridge dark
No gold accent barSlide titles lack signature element::after pseudo-element, 40px × 3px, #b79d64
Missing mental modelSlide 3 absentTwo-column comparison + dark insight callout
Cormorant 600/700 loadedFont URL includes bold weightsRemove from Google Fonts URL — max 500
Service items wrappingLabels too longKeep under ~35 characters
Copy button gold on creamCopied state invisibleDark bg #1a1a1a + gold text #b79d64