Seven variants, one spacing constant, and a colour contract that means the system
ships one file per variant instead of three. Every measurement on this page was taken
from the artwork, not estimated from it.
Status
Approved · Canonical · 2026-07-25
Variants
7 approved · plus 2 fixed colourways
Canonical source
foundations/identity.md
Open items
4 of 7 resolved · 3 excluded by name
Approved: the first canonical release
Fernando Kylas approved the Identity System on 2026-07-25. The approval covers
the complete identity architecture and the philosophy that governs it: all seven variants, both fixed colourways, the mark's rationale, the currentColor contract, the
construction rules and the misuse boundaries. Three operational follow-ups were acknowledged and
intentionally excluded: LOGO-02, LOGO-04 and LOGO-07 (chapter 08).
No undocumented change may modify the approved identity; a change supersedes a
decision in DECISIONS.md, with a reason. Chapter 10 explains why.
The artwork on this page is live
Every logo below is the production SVG from assets/logos/export/, injected as an
inline sprite and driven by currentColor, not a screenshot.
Toggle Theme and watch every appearance re-colour itself from one set
of files. That is the pipeline, demonstrated rather than described. The markdown at
foundations/identity.md remains authoritative for the rules.
Chapter 01
Identity overview
The identity has three elements and no more. Everything else in the system is typography,
rule and space.
The mark
A six-armed asterisk. Radially symmetrical about
its centre.
The wordmark
A single lowercase word, no space, in a heavy
geometric sans.
The descriptor
Set in Strawford. Present in one
lockup only, with its own measured spacing.
The most important usage rule in this chapter
The logo appears less often here than in most brands. One appearance per surface is the
default.
A consultancy that publishes evidence earns attention through the substance of what it
publishes. A logo repeated on every page, in every corner, at every scale reads as a brand asserting itself, which is exactly the posture this identity avoids.
The corollary matters more than the rule: where the logo is absent, typography must
be doing its job. A page that needs a logo to look like Fernando Kylas has a
typography problem, not a branding problem.
Surface
Logo appears
A report
Once, on the cover. Not on interior pages
A page of a document
In the running header only if the document is loose-leaf
A website page
Once, in the header
A proposal
Once, on the cover
A slide deck
On the opening and closing slides only
A social post
Once, or not at all where the account itself is the attribution
Why the mark exists · confirmed by Fernando Kylas, 2026-07-25
The mark carries two meanings. They are not two stories.
The first is personal. Long before it was a logo, the star stood for hope
during a difficult period in Fernando's family. His birth marked a new beginning after
loss, and the star became a quiet reminder that light can emerge from difficult moments.
It is not a symbol of fame or recognition. It stands for optimism, resilience and new
beginnings.
The second is professional. As the work moved into consulting, the same
symbol came to sit alongside the asterisk used throughout publishing, research, technical
documentation and professional writing. An asterisk tells a reader that there is
additional context, supporting evidence or reasoning worth exploring. That is how this
practice works with clients: not simply delivering recommendations, but explaining the
thinking behind them.
Together they describe one philosophy. The mark stands for clarity emerging from uncertainty: personally, hope; professionally, understanding.
This rationale was not written to justify the design after the fact. The symbol
accumulated these meanings over many years of personal life and professional practice;
the logo came afterwards.
What this settles
Every future extension of the identity reasons from here: the mark may be quiet, small or
abstract, but it may never be decorative. It already means something.
Chapter 02
Logo architecture
Seven variants, each with a job. A variant is chosen by the space available and the context's
need for the full name, never by preference. Read the cards for the decision each
one answers, not for the artwork it contains.
The default, and the one to reach for when you are not sure.
Symbol and name together, in the proportion the places a logo actually goes already have.
Strengths
Complete; nothing has to be inferred. Its 7.573:1 ratio suits website headers, document headers, slide footers and email headers without being cropped or padded to fit.
Use for
Website and documentation headers, client portals, application chrome.
Avoid
Below 120 px wide, where the wordmark closes up. That is what the abbreviated horizontal is for.
2.551 : 1min 120 pxfk-logo-stacked.svg
Stacked
Chosen when the composition has a centre to hold rather than an edge to sit on.
Strengths
A near-square footprint, so it centres well and holds the middle of a composition instead of sitting at its edge.
Use for
Report and proposal covers, title slides, portrait social, merchandise.
Avoid
In a horizontal band. It wastes height and forces the wordmark smaller than the horizontal lockup would at the same header height.
1.998 : 1min 180 pxfk-logo-stacked-descriptor.svg
Stacked with descriptor
The only variant that says what the practice does, which is why it belongs to first contact and nowhere else.
Strengths
Answers “who is this and why should I read on?” in one object, without a line of supporting copy.
Use for
First contact: a proposal cover for a new client, a conference slide, a printed introduction, a press kit.
Typeface
The descriptor is set in Strawford, the one place both brand voices meet inside a single artwork. The wordmark's heavy geometric sans carries the name; Strawford carries the line that explains it.
Avoid
Where the audience already knows the practice; the descriptor becomes noise. Never below 180 px, because the descriptor fails first, before anything else in the lockup. Never re-typed as live text: see chapter 03.
6.546 : 1min 80 pxfk-logo-wordmark.svg
Wordmark
For places already attributed, where a mark would only repeat what the page has said.
Strengths
Says the name, which the mark cannot. The single-word compression makes a thirteen-character name behave as one object.
Use for
An email signature, a document footer, a letterhead.
Avoid
Where the horizontal lockup fits. The mark earns its place whenever there is room for it.
The primary lockup's structure at a quarter of its width. A constraint response, not a style choice.
Strengths
Keeps symbol-plus-name intact where the full lockup would have to be shrunk below its minimum.
Use for
Mobile headers, narrow sidebars, compact application chrome.
Avoid
On desktop, or anywhere the full horizontal lockup fits. Reaching for it by preference is how a system loses its primary lockup.
1 : 1min 16 pxfk-logo-mark.svg
Mark
The only form that survives 16 px and a circular crop, and the only one with no name in it.
Strengths
Square canvas, so it drops into a favicon, app icon, avatar or stamp without adjustment. Radially symmetrical, so a circular crop takes nothing from it. Carries no language, so it needs no localisation.
Use for
Favicon, app icon, social avatar, a repeated small-format stamp.
Avoid
As the primary identifier on a first encounter. Someone who has never seen the brand learns nothing from it: the mark identifies, it does not introduce.
The smallest form that still carries a name, and the reason it is never paired with the mark.
Strengths
Nearly square, so it behaves like a monogram where the mark would read as too abstract for the context.
Use for
Compact avatars, a favicon alternative, a mobile header too narrow for even the abbreviated lockup.
Avoid
Any first-contact context, and anywhere the mark is also present, because the asterisk plus fk, without the full name, is two abbreviations and no identity.
Choosing a variant
Ask the questions in order and stop at the first yes. Every question is about the
context, never about preference, which is the point: a variant is a consequence of
where it lands, not a choice made in advance.
Ask
If yes
Does the reader need to know what the practice does?
stacked-descriptor · min 180 px
Does the composition have a centre to hold?
stacked · min 120 px
Is the page already attributed, so a mark would repeat it?
wordmark · min 80 px
Is there 120 px of width?
horizontal, the default
Is there at least 48 px?
horizontal-abbreviated
Is the space square, and is the mark legible in the context?
mark · min 16 px
Square and small, but the mark reads as too abstract?
wordmark-abbreviated · min 24 px
If none of them is a yes, the space is too small for the identity, and the correct answer
is to place no logo at all rather than one below its minimum.
Chapter 03
Construction
Measuring the lockups revealed a single spacing constant they all share, and one master with a real defect that had survived every review because it was 0.43 units wide. Both are
arguments for measuring artwork rather than trusting it.
Corrected in the pipeline
The mark's artwork overflowed its own viewBox
The master declared viewBox="0 0 155.35 155.35", but its
artwork spans x = -0.43 to 155.79. It overflowed its own
canvas by 0.43 units on the left and 0.44 on the right, so
both outer arm tips were clipped by any renderer honouring the viewBox.
At favicon size the loss is about 0.04 px per side: invisible, and therefore exactly the
kind of defect that survives review indefinitely. It also made a nominally square file
1.0055:1.
The correction is -0.43 -0.43 156.22 156.22: the true artwork
width, squared, with the artwork centred. Nothing is clipped and the canvas is exactly 1:1, which is what makes the favicon and app-icon applications work.
Accepted into the master on 2026-07-25. The file now declares this viewBox
itself and is the authoritative source; the pipeline's correction for the mark has been
removed. The export is byte-identical either way, verified by hashing all
three mark exports before and after, which is what made the handover safe to make.
Variant
Master viewBox
Overflow
Corrected to
Corrected where
mark
0 0 155.35 155.35
0.43 left · 0.44 right · 0.01 bottom
-0.43 -0.43 156.22 156.22
the master
stacked
0 0 1440 564.48
0.0099 right
0 0 1440.01 564.48
pipeline
stacked-descriptor
0 0 1440 720.59
0.0099 right
0 0 1440.01 720.59
pipeline
horizontal-abbreviated
0 0 373.14 155.77
0.01 bottom
0 0 373.14 155.78
pipeline
horizontal · wordmark · wordmark-abbreviated
—
none
unchanged
—
Pipeline first, master second
Corrections land in the pipeline before they land in a master, because the masters are
Fernando's artwork and a correction is a proposal until he accepts it. Chapter 07 explains why
that ordering matters.
Only the mark's viewBox attribute was edited: no path, no id, nothing else.
Path data is passed through byte-for-byte, verified identical across all seven
variants. The three remaining corrections are rounding artefacts of 0.01 units or
less and are best folded in whenever those masters are next re-exported from Illustrator.
The lockups share one spacing constant
Measuring the gap between the mark and the type in each lockup gives the same answer three
times.
Lockup
Mark height
Gap
Gap ÷ mark height
horizontal
155.76
38.64
24.81%
horizontal-abbreviated
155.36
40.42
26.02%
stacked
275.02
69.47
25.26%
The unit
The lockups are built on one quarter of the mark's height. Call that unit s.
In stacked-descriptor, the wordmark-to-descriptor gap measures
63.81 against s = 68.76, close to s, and slightly tighter, which is correct: a descriptor binds to the name
above it more closely than the name binds to the mark.
This is a discovered constant, not an imposed one. It is recorded because
every future lockup, extension or application must use it, and because it gives clear space
a principled basis rather than an arbitrary one.
The descriptor, for a future live-text extension
The descriptor ships as outlined artwork, so nothing currently needs these values. They are
recorded because the moment anyone sets that line as live text, the artwork is the specification, and eyeballing it will not reproduce it.
Property
Measured
As a ratio
Typeface
Strawford
confirmed 2026-07-25
Advance width
1020.89
0.7089 of the lockup width
x span
209.55 to 1230.44
centred on the lockup
Band height
92.30
—
Ascender height
69.66
—
x-height
46.25
0.6640 of the ascender
Gap from the wordmark
63.81
0.928 of s
Weight, optical size and tracking are not specified, and must not be guessed.
Three independent attempts to derive them from browser text metrics disagreed with each
other, because canvas metrics, DOM range rectangles and CSS weight resolution each measure
something subtly different. Publishing a figure from any one of them would be exactly the
false precision this system refuses elsewhere.
What is established: the letterforms are unambiguously Strawford: the double-storey
a, the single-storey g with an open tail, the angled t apex, and the artwork's ascender-to-x-height ratio matches Strawford's while disagreeing
with Geomanist's by roughly 8%. Match the remaining three values optically against
the outlined artwork at a large size, then record them here.
Clear space
Minimum clear space on all four sides is 2s, half the mark's height.
Double the lockup's own internal gap, so external breathing always exceeds internal spacing.
Below that, the mark starts to belong to whatever it sits next to.
The keep-out zone
the solid rule is
the artwork's own bounds ·
the dashed rule is the
keep-out boundary at 2s. Nothing may enter the band between them.
In code
/* primary lockup, sized by width */
.fk-logo--horizontal {
inline-size: 180px;
padding: 5.4%; /* 2s against the width */
box-sizing: content-box;
}
Clear space is a layout rule applied at use, never baked into the file.
Every export is tight-fitted precisely so clear space stays the consuming layout's
decision.
Variant
s
Clear space 2s
% of rendered width
% of rendered height
mark
38.84
77.68
49.7%
49.7%
wordmark
28.98
57.96
4.7%
30.5%
wordmark-abbreviated
38.02
76.04
43.1%
50.0%
horizontal
38.94
77.88
5.4%
41.0%
horizontal-abbreviated
38.84
77.68
20.8%
49.9%
stacked
68.76
137.51
9.5%
24.4%
stacked-descriptor
68.76
137.51
9.5%
19.1%
For the wordmark-only variants there is no mark, so s is keyed to the type: one
quarter of the wordmark's x-height (115.91 units) for wordmark, and one quarter
of the glyph height for wordmark-abbreviated. The one exception to tight
fitting is mark, whose square canvas carries 0.43 units (0.28%) of vertical padding by design, so the favicon canvas stays exactly 1:1.
Minimum size
Judged from a rendered size ladder, not calculated. The minimum is the
smallest width at which every element remains separable and the type does not close up. The
ladder below is live artwork at the stated widths; the leftmost in each row is the minimum.
Mark · minimum 16 px, limited by arm separation
16 px min24 px32 px48 px
Horizontal · minimum 120 px, limited by wordmark letterfit
120 px min180 px240 px
Stacked with descriptor · minimum 180 px, limited by the descriptor
120 px: too small180 px min280 px
Variant
Minimum screen width
Limiting element
Print minimum
mark
16 px
Arm separation
5 mm
wordmark-abbreviated
24 px
f / k counters
7 mm
horizontal-abbreviated
48 px
f / k counters
13 mm
wordmark
80 px
Letterfit across 13 characters
22 mm
horizontal
120 px
Wordmark letterfit
32 mm
stacked
120 px
Wordmark letterfit
32 mm
stacked-descriptor
180 px
The descriptor
48 mm
Print minima are conversions, not proofs
Derived at 96 dpi from the screen minimum, which is conservative, because a 300 dpi press holds finer detail than a screen, so the true print minimum is probably smaller.
Confirm on a physical proof before any press run.
Optical alignment
Four rules, and the first two exist because the artwork is already correct: the temptation
is to tidy it.
The mark is optically centred, not mathematically centredartwork is 156.22 × 155.36, 0.86 units wider than tall. Do not "correct" it; that is how the arms balance
Align lockups on the wordmark's baseline, never on the bounding boxthe f ascender and the y descender both extend past the x-height band. Baseline sits at 155.23 of 190.15
Never re-space a lockupthe mark-to-type gap is s; changing it breaks the constant
Never rebuild a lockup from partsplacing the mark beside the wordmark and eyeballing the gap will not reproduce the lockup file
Chapter 04
Colour behaviour
Production SVGs carry fill="currentColor". There is no black file, no white file
and no green file for internal use; the consuming application supplies the colour through
CSS.
The production contract
<!-- one file. the colour comes from CSS. -->
<span class="fk-logo" style="color: var(--fk-text-primary)">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 1440 190.15"
fill="currentColor" role="img" aria-label="Fernando Kylas" focusable="false">
<path d="…"/>
</svg>
</span>
.fk-logo svg { display: block; inline-size: 100%; block-size: auto; }
/* inherits the theme automatically; no second asset, no swap logic */
.fk-logo { color: var(--fk-text-primary); }
Why this and not three colourways
currentColor
Three fixed files
One file per variant, 7 files
7 × 3 = 21 files to keep in sync
Theme inversion is free
Needs swap logic in every consuming product
Cannot drift from the palette
Drifts the moment a token changes
Correct under prefers-contrast, forced colours, print
Each mode needs its own asset
Inherits the colour language by construction
Repeats a hex the palette already owns
The Report Engine's dark theme is a single attribute. A logo that inherits
currentColor follows it without being told; a logo that hardcodes
#0A0A0A becomes invisible. The contract is not really about file count: it is about which side of the boundary the colour decision lives on.
Put it in the file and every consumer inherits a decision it cannot revisit.
One file, every permitted colour
Every appearance below is the same fk-logo-horizontal.svg, re-coloured by CSS
alone. Ratios are the logo as a graphical object against its ground: 3:1 minimum.
Light paper · text-primary
18.03:1 · permitted
Light paper · text-secondary
8.07:1 · a deliberately quiet placement
Light paper · accent
2.65:1 · never
Dark ink · text-primary
19.04:1 · permitted
Dark raised · text-primary
13.73:1 · permitted
Dark ink · accent
6.86:1 · legal, still discouraged
The logo is never green on a light ground
At 2.65:1 it fails the 3:1 graphical-object threshold. This is the single
most likely misuse in the system, because green is the brand's accent and the logo is the brand's mark, and the instinct to combine them is strong and wrong.
On a dark ground green is legal at 6.86:1 but still discouraged: the logo is
an identifier, not a status signal, and green in this system means accent or
positive finding. A green logo says something the logo should not be saying. Use it
only where Fernando has approved the specific application.
Fixed-colour exports: external use only
Use fixed-colour files for
Print and press
A press kit
Office documents: Word, PowerPoint
Third-party and partner usage
Event and sponsor artwork
Email signatures
Anywhere a file leaves Fernando's control
Never use them for
The website
The Report Engine
The Documentation Engine
The Publishing Platform
Any product that can inherit a CSS colour
A fixed file inside a themed product is a latent bug: correct in light mode,
invisible in dark.
Colourway
Value
Token
For
black
#0A0A0A
text-primary light
Light and white grounds
white
#FAFAF7
text-primary dark
Dark and photographic grounds
Two colourways, not three
There is no green fixed export, because green fails on light grounds
and shipping a conditionally-legal file to an external party guarantees it will eventually
appear on a white page.
Neither colourway is pure
#0A0A0A not
#000000; #FAFAF7 not
#FFFFFF. The reasons hold for a logo as much as for text; see the Colour Language.
Fixed-colour exports omit the accessibility attributes: they are placed by
a human into a document that supplies its own context, and a stale
aria-label inside a Word file helps nobody.
Chapter 05
Incorrect usage
Each of these is a real failure mode with a specific consequence, not a generic prohibition.
Every example below is rendered wrong on purpose, because a rule you have seen broken is a
rule you remember, and this identity has no decoration to hide a mistake behind.
Never
Stretch. The mark's radial symmetry and the wordmark's
letterfit both distort, and a geometric sans makes it obvious. Scale uniformly.
Never
Rotate. The asterisk is radially symmetrical, so
rotation reads as a mistake rather than a choice.
Never
Reduce opacity. It silently drops measured contrast: 40% ink on paper is roughly 1.9:1. To quiet the logo, use
text-secondary at 8.07:1.
Never
Green on a light ground.2.65:1 fails the 3:1 graphical-object threshold. Use
text-primary at 18.03:1.
Never
Add a shadow, glow or bevel. The system has one shadow
token and strips it in print. A shadowed logo is unprintable and dates instantly.
Never
Outline or stroke. The artwork is filled shapes; a
stroke adds weight the design does not have. Fill only.
Never
Place on a busy or photographic ground. Contrast
becomes unmeasurable and varies across the artwork. Use a solid token ground, or a solid
panel behind the logo.
Never
Crop or bleed. The mark is a complete symbol; a cropped
asterisk reads as damage. Full artwork inside its clear space.
Never
Rebuild a lockup from parts. Eyeballing the gap will not reproduce the lockup file; the real gap is s, one quarter of the mark's height.
Use fk-logo-horizontal.svg.
The full list
The nine above are the ones worth seeing. These are the rest, each with the specific thing it breaks, because a prohibition a reader cannot explain is a prohibition they will
eventually talk themselves out of.
Misuse
What breaks
Instead
Stretching, non-uniform scale
Radial symmetry and letterfit both distort
Scale uniformly · block-size: auto
Re-spacing a lockup
Breaks the s constant; no longer matches any other appearance
Use the lockup file unmodified
Rebuilding a lockup from parts
As above, plus baseline drift
Use the lockup file
Random colours
Contradicts a three-hue palette and usually fails 3:1
text-primary
Green on a light ground
2.65:1, failing the graphical-object threshold
text-primary at 18.03:1
Drop shadows, glows, bevels
Unprintable; the system strips shadow in print
Nothing. Contrast, not effects
Rotation
Reads as a mistake, given radial symmetry
Never rotate
Opacity below 100%
Silently drops measured contrast
Full opacity · text-secondary to quiet it
Busy or photographic backgrounds
Contrast becomes unmeasurable
A solid token ground, or a solid panel
Below minimum size
Arms merge; counters close
The next variant down
Outlining or stroking
Adds weight the design does not have
Fill only
Gradient fill
The palette has no gradients
Solid currentColor
Cropping or partial bleed
A cropped asterisk reads as damage
Full artwork inside its clear space
The mark plus fk, without the full name
Two abbreviations, no identity
horizontal-abbreviated is the sanctioned pairing
A hand-typed tagline under the stacked lockup
Will not match the descriptor's measured spacing
stacked-descriptor
Reordering or recolouring a single arm
It is one path; there are no parts
Never
A fixed-colour file inside a themed product
Correct in light, invisible in dark
The currentColor export
Baking clear space into the file
Every consumer inherits padding it cannot remove
Tight-fitted file, padding at use
Editing a file in export/
Overwritten on the next pipeline run
Fix the master, re-run the pipeline
Chapter 06
Applications
The mockups below are schematic on purpose. They show the decision: which variant, at what size, and what made that the answer, rather than a pretend screenshot of a product
that does not exist yet.
Website header, desktop. 130 px of width clears the 120 px minimum, so
the primary lockup is available and therefore correct. currentColor means the
same file serves the dark theme. Once, top-left, linked to the root.
Website header, mobile. 56 px is under horizontal's 120 px
minimum and over this variant's 48 px, so the constraint chose the variant. Shrinking the
primary lockup to fit would have been the same decision made badly.
Digital Visibility Diagnostic
Report cover. A cover has a centre to hold, which is the question
stacked answers. Once here and never on an interior page: the
report's title carries those, and a logo repeated forty times stops being a signature.
Presentation, opening slide. First contact: the room may not know what
the practice does, so the descriptor earns its place here and loses it on slide two. At
180 px it is exactly at its minimum; below that the descriptor fails before the wordmark
does. Reversed on ink at 19.04:1.
16 px
32 px
avatar
app icon
Favicon, avatar, app icon. Three square spaces, one answer. The
corrected square viewBox is what removes the need for a separate icon file, and radial
symmetry is what survives a platform cropping to a circle without being asked.
Fernando Kylas
Digital impact consultant
fernandokylas.com
Email signature. The name is already in the signature block, so a mark would repeat it; wordmark at 110 px. And email clients cannot reliably inherit
CSS, so this is one of the few places a fixed-colour file is the right
answer rather than a shortcut.
Every application, decided
Every surface the identity currently reaches, with the decision already made. Where the
variant column reads none, that is the decision, not an omission.
Application
Variant
Colour
Notes
Report cover
stacked
text-primary
Once, on the cover. Never on interior pages
Report running header
none
—
The report title carries the header
Documentation site header
horizontal
inherits theme
Minimum 120 px
Website header, desktop
horizontal
currentColor
Inherits light and dark automatically
Website header, mobile
horizontal-abbreviated
currentColor
Minimum 48 px
Client portal
horizontal
currentColor
Once, top-left, linked to the portal root
PDF
stacked on the cover
text-primary
Print forces the light theme, so the logo renders ink
Presentation, opening
stacked-descriptor
as ground requires
Minimum 180 px. First contact
Presentation, closing
stacked
as ground requires
No logo on interior slides
Email signature
wordmark
black fixed
Email cannot inherit CSS reliably
Favicon
mark
fixed, both
Ship light and dark via prefers-color-scheme
App / touch icon
mark
fixed
The square canvas is why the mark exists in this form
Social avatar
mark or wordmark-abbreviated
fixed
Platforms crop to a circle; radial symmetry survives it
Social post
usually none
—
The account is the attribution
Proposal cover
stacked-descriptor
text-primary
First contact
Invoice / letterhead
wordmark
black fixed
A mark is redundant where the document is already attributed
Press kit
all seven
fixed, both
The only place the full set ships together
Merchandise, signage
stacked or mark
as substrate requires
Confirm on a physical proof
Favicon: the one sanctioned exception to "no colour inside a logo file"
A favicon has no CSS context to inherit from, so a single SVG may adapt to the browser's
theme with an embedded media query.
Why the mark exists at all, given typography carries the identity.
Chapter 01 records what the mark means; this is the separate question of why the
system carries one at all, given the Colour Language argues the identity lives in the type.
The answer is the sizes and shapes typography cannot serve: a 16 px favicon, a circular
avatar, a square app icon. fernandokylas at 16 px wide is illegible; the mark
at 16 px is not.
The mark is a wayfinding device for small and square spaces, not the identity itself, which is why it is never the primary identifier on first contact.
Decision 02
Why the wordmark is one lowercase word.
One object, not two words.fernandokylas behaves as a
single shape, which is why the horizontal lockup holds together at 120 px where "Fernando
Kylas" would break into two.
Lowercase is quieter. Capitals assert; lowercase states. The
identity's stated qualities are quietly confident, and the wordmark is where that is literally set in type.
It matches the URL. The name a reader types and the name they see are
the same string.
Decision 03
Why restraint matters here specifically.
Every element of this identity is load-bearing, so there is nothing to hide behind. A
system with a pattern library, three accent colours and an illustration style can absorb a
weak logo. This one cannot: if the wordmark is stretched or the mark is clipped, there is
no decoration to distract from it.
Restraint raises the cost of every error, which is the argument for documenting
the errors. Hence chapter 05.
Decision 04
Why the export pipeline is a script and not a manual export.
Seven variants times three outputs is 21 files, each needing identical cleaning, the same
viewBox corrections and the same accessibility attributes. Done by hand, they drift, which is exactly what happened in Figma, where two colour styles sat one hex digit from
canon for long enough to pass review.
build-exports.py --check makes drift a failure rather than a
discovery.
Decision 05
Why corrections go to the pipeline first, and the master second.
assets/logos/source/ holds Fernando's authored artwork, so a correction to it
is a proposal until he accepts it. The mark's viewBox defect was therefore fixed in the pipeline first, and that ordering is what made the fix safe to hand over.
By the time he accepted it on 2026-07-25, the corrected output had already been in use and
verified, so applying it to the source changed nothing at all.
A correction proved before it is applied is a correction that cannot surprise anyone. The
three remaining rounding artefacts follow the same route.
Chapter 08
Open items
Seven items were opened with this chapter. Four are closed. The three that
remain were acknowledged in Fernando's approval and intentionally excluded from it: each is an operational follow-up, and none can change the artwork, the architecture
or the rules. They are kept together, resolved alongside open, so a reader can see what the
chapter did not know and when that changed.
LOGO-01
The mark's meaning is documented
Resolved 2026-07-25
Previously the chapter's one blocking item: no rationale for the mark
existed in this repository, in Figma, or in Notion, and chapter 01 could only offer a
reading labelled Exploratory.
Fernando confirmed the rationale on 2026-07-25 and it
is now canonical in chapter 01: the star as hope, the asterisk as understanding, one
philosophy of clarity emerging from uncertainty. The exploratory reading has been replaced,
not amended. Recorded in DECISIONS.md.
Retained here for provenance rather than deleted, so a reader can see
what the chapter did not know and when that changed.
LOGO-02
Print minima are unproofed
Medium
Chapter 03's print column is converted from screen minima at 96 dpi,
which is conservative. A physical proof would likely allow smaller.
Needed before any press run.
Acknowledged and excluded from the approval. It cannot
invalidate an approved rule: a proof can only ever permit a smaller size than the
conservative figure already documented. If it does, chapter 03 is relaxed by a superseding
decision.
LOGO-03
The mark's master is the authoritative source
Resolved 2026-07-25
The correction was accepted into
Fernando-Kylas_Logo - Mark & Favicon.svg, which now declares
viewBox="-0.43 -0.43 156.22 156.22" itself. The pipeline's
correction for the mark has been removed.
All three mark exports hash identically before and after,
so accepting the fix into the source changed no output; the proposal had been proved before
it was applied.
Deliberately still in the pipeline: the three sub-hundredth rounding
corrections for stacked, stacked-descriptor and
horizontal-abbreviated. They are Illustrator artefacts of 0.01 units or less,
not defects, and are best folded in when those masters are next re-exported.
LOGO-04
No PNG rasters are produced
Low
Every current consumer takes SVG. Apple touch icons and some social
platforms will need rasters.
Deferred until a real consumer exists, rather than
generating 42 unused files.
Acknowledged and excluded from the approval. A raster
derived from an approved master inherits its approval; generating one is production work, not
a design decision.
LOGO-05
The identity is approved and canonical
Resolved 2026-07-25
Previously: a file existing in export/ is not approval, and
neither is a frame existing in Figma.
Fernando Kylas approved the Identity System on 2026-07-25
as the first canonical release. All seven variants and both fixed colourways are Approved, together
with the philosophy that governs them. Recorded in DECISIONS.md.
This was the one item that could never be closed by further work. Worth
naming what changed: assets/logos/export/ is now canonical approved output rather
than derived-and-pending. The pipeline's role is unchanged: export/ is still generated, still never edited by hand, and --check still fails on drift.
What changed is that drift is now a governance breach rather than a stale
file.
LOGO-06
The descriptor is set in Strawford
Resolved 2026-07-25
Confirmed by Fernando. The previous entry guessed
Geomanist at a lighter weight; that guess was wrong and is superseded.
Independent checks agree with the confirmation: the letterforms match Strawford's
a, g and t, and the ascender-to-x-height ratio matches
Strawford while disagreeing with Geomanist by roughly 8%.
The residual task this entry once carried is now tracked separately as
LOGO-07, so that each item excluded from the approval has an addressable
ID.
LOGO-07
The descriptor's weight, size and tracking are unspecified
Low
The typeface is confirmed; the three setting values are not. Three
attempts to derive them from browser text metrics (canvas metrics, DOM range rectangles, and
CSS weight resolution across the three declared Strawford faces) disagreed with each other, so no figure has been published. Publishing the best-agreeing number would be
false precision in a chapter where every other value was measured.
Chapter 03 instead records the artwork geometry a live-text setting must
reproduce, and requires the three values to be matched optically against the outlined
artwork before any live-text extension ships.
Acknowledged and excluded from the approval. Needed only
if the descriptor is ever set as live text. Nothing currently requires it; it ships as outlined artwork inside one lockup, and the artwork remains the specification either way.
Chapter 09
Identity evolution
This is not a history of the artwork. It is a record of how the meaning accumulated, because the symbol acquired its meanings before it acquired a specification, and a reader who
knows only the current state will eventually mistake accumulation for design.
Most identities are drawn and then explained. This one ran the other way: the symbol meant
something for years before anyone asked it to work at 16 px. Four stages, each of which added
a meaning without displacing the one before it.
01
A personal symbol
Meaning: hope
The star stood for hope during a difficult period in Fernando's family, long before it was
anything else. It carried no professional weight and was not intended to. What it
established is the thing every later stage inherited: the symbol was never
decorative, because it was never chosen for how it looked.
02
A professional symbol
Meaning: understanding
As the work moved into consulting, the same shape came to sit alongside the asterisk of
publishing, research and technical documentation: the mark that tells a reader there is
more to know. Nothing was redrawn. A second meaning arrived because the context
changed, not because the symbol did.
03
A unified identity
Meaning: both, as one
The two meanings turned out to be one. Hope is clarity arriving in a personal life;
understanding is clarity arriving in a client's. Recognising that they were the same
philosophy is what made the symbol usable as an identity rather than as two coincidences, and it is why chapter 01 refuses to present them as separate stories.
04
A canonical Brand System
Meaning: inheritable
The final stage added no meaning at all. It made the existing meaning transferable:
measured, documented, versioned, and answerable by someone who was not there. Until
2026-07-25 this chapter could describe how the mark behaves but not why it exists, and a
system that cannot answer why cannot survive the person who built it.
What the sequence teaches
Meaning accumulated; it was never assigned. That is why the rationale is credible, and why the next extension of this identity has to earn its meaning the same way rather than being
given one.
Chapter 10
Identity stewardship
From here the identity is governed by documented principles rather than personal preference.
That is a deliberate transfer of authority, and it changes what a future disagreement looks
like.
The rationale is canonicalwhat the mark means is written down, not remembered
Decisions are preserved, not overwrittenthe exploratory reading this chapter once carried is still on record, marked superseded
A change supersedes; it does not replacea new decision cites the one it retires and says why
The identity evolves through governance, not opiniona proposal is answered from the documentation, or it changes the documentation
The practical effect is small and permanent. A proposal to use the mark as a pattern, a
texture or a repeated background is now answerable from chapter 01 rather than from taste; the mark already means something, so it cannot be decorative. If that constraint is ever
wrong, the way to change it is to supersede the decision in DECISIONS.md with a
reason, not to quietly do something different.
Nothing here removes Fernando's authority; he remains the only approver. What it removes is
the need for him to be present for the identity to be applied correctly.
The point of writing it down
A brand system earns its name at the moment it stops needing its author.
Every measurement in this chapter was taken rather than estimated, every prohibition carries
the consequence that justifies it, and every open question is recorded as a question rather
than smoothed over. That is not thoroughness for its own sake. It is what allows someone
arriving in five years to apply this identity correctly, disagree with it precisely, and
change it accountably.
Seven variants. One spacing constant. One colour, supplied by whatever the logo lands in.
Fernando Kylas
Provenance
Canonical source.foundations/identity.md, authoritative for every rule on this page.
Artwork. Seven SVG masters in assets/logos/source/, authored by Fernando in Illustrator. Six were read read-only and are unmodified; one was edited with authorisation. See Master edited below. The artwork on this page is injected from assets/logos/export/ by docs/identity/build-sprite.py.
Measurements. Artwork bounds, lockup gaps and the s constant measured 2026-07-25 using a browser's getBBox() against the masters and every export. Path data verified identical between master and export for all seven variants.
Minimum sizes. Judged from a rendered size ladder at 1× on #F6F4F1, 2026-07-25. A judgement, not a measurement.
Figma. Page 04 — Logo System in file 8THxEHQJfyzTAdJbd83YcO. Audited 2026-07-25.
Rationale. The mark's meaning (chapter 01) was confirmed by Fernando Kylas on 2026-07-25 and is canonical. It closed LOGO-01, the chapter's only blocking item.
Descriptor. Set in Strawford, confirmed 2026-07-25 (LOGO-06). The previous observation-from-rendering, Geomanist at a lighter weight, was wrong and is superseded. Weight, optical size and tracking remain unspecified (LOGO-07); chapter 03 records why and what a live-text extension must reproduce.
Master edited.Fernando-Kylas_Logo - Mark & Favicon.svg: the viewBox attribute only, on 2026-07-25, with Fernando's explicit authorisation (LOGO-03). All three mark exports hash identically before and after.
Approval. Approved and canonical. Approved by Fernando Kylas on 2026-07-25 as the first canonical release, covering all seven variants, both fixed colourways and the governing philosophy. LOGO-02, LOGO-04 and LOGO-07 were acknowledged and intentionally excluded. Recorded in DECISIONS.md; this closed LOGO-05.
Release designation.The first canonical release, in Fernando's own words. No formal release-numbering scheme has been decided; ROADMAP.md carries Versioning and release process in its backlog. foundations/identity.md is at document version 4, which is that chapter's revision count and not a release number.
Scope. This approval covers the Identity System and nothing else. Colour Language and foundations/typography.md remain Ready for review and not approved, the paper tone excepted.
Related.Colour Language · assets/logos/README.md · foundations/typography.md