The code and data behind xeiaso.net
feat: add ai-writing-tropes skill from tropes.fyi (#1164)
Comprehensive catalog of AI writing patterns to avoid, organized into
six categories: word choice, sentence structure, paragraph structure,
tone, formatting, and composition. Each category has its own reference
file with examples and fixes.
https://claude.ai/code/session_01S1R6eHV1ftixGbx4sWGg8N
Co-authored-by: Claude <noreply@anthropic.com>
+612
+94
.claude/skills/ai-writing-tropes/SKILL.md
+94
.claude/skills/ai-writing-tropes/SKILL.md
···
1
+
---
2
+
name: ai-writing-tropes
3
+
description:
4
+
Detect and eliminate common AI writing tropes from prose. Use when drafting,
5
+
editing, or reviewing text to avoid the predictable patterns that mark
6
+
AI-generated writing. Source: tropes.fyi
7
+
metadata:
8
+
trigger:
9
+
Writing prose, editing drafts, reviewing content for AI tells, or when user
10
+
mentions "tropes", "AI patterns", "slop", or "tropes.fyi"
11
+
author: tropes.fyi
12
+
---
13
+
14
+
# AI Writing Tropes to Avoid
15
+
16
+
Comprehensive catalog of AI writing patterns that make text feel machine-generated.
17
+
Any single pattern used once might be fine. The problem is when multiple tropes
18
+
appear together or when one trope repeats throughout a piece.
19
+
20
+
## How to Use
21
+
22
+
When writing or editing prose:
23
+
24
+
1. Draft the content.
25
+
2. Check against the trope categories below.
26
+
3. If you spot a pattern, rewrite that passage.
27
+
4. Re-read the whole piece for pattern density — a few is tolerable, a cluster is
28
+
not.
29
+
30
+
## Trope Categories
31
+
32
+
### Word Choice
33
+
34
+
Overused vocabulary and phrasing that AI defaults to. See
35
+
[references/word-choice.md](references/word-choice.md).
36
+
37
+
Key offenders: "quietly", "delve", "tapestry", "landscape", "serves as",
38
+
"leverage", "robust", "harness", "streamline".
39
+
40
+
### Sentence Structure
41
+
42
+
Formulaic sentence patterns that no human writes at scale. See
43
+
[references/sentence-structure.md](references/sentence-structure.md).
44
+
45
+
Key offenders: negative parallelism ("not X — it's Y"), dramatic countdowns
46
+
("Not X. Not Y. Just Z."), self-posed rhetorical questions, anaphora abuse,
47
+
tricolon abuse, gerund fragment litanies.
48
+
49
+
### Paragraph Structure
50
+
51
+
Layout and organization patterns that betray AI generation. See
52
+
[references/paragraph-structure.md](references/paragraph-structure.md).
53
+
54
+
Key offenders: short punchy fragments as standalone paragraphs, listicles
55
+
disguised as prose.
56
+
57
+
### Tone
58
+
59
+
Voice and framing habits that sound performative. See
60
+
[references/tone.md](references/tone.md).
61
+
62
+
Key offenders: false suspense ("Here's the kicker"), patronizing analogies
63
+
("Think of it as..."), false vulnerability, stakes inflation, vague attributions,
64
+
invented concept labels.
65
+
66
+
### Formatting
67
+
68
+
Visual and typographic tells. See
69
+
[references/formatting.md](references/formatting.md).
70
+
71
+
Key offenders: em-dash addiction, bold-first bullets, unicode decoration.
72
+
73
+
### Composition
74
+
75
+
Document-level structural problems. See
76
+
[references/composition.md](references/composition.md).
77
+
78
+
Key offenders: fractal summaries, dead metaphors beaten into the ground,
79
+
historical analogy stacking, one-point dilution, signposted conclusions.
80
+
81
+
## Quick Self-Check
82
+
83
+
Before delivering prose, ask:
84
+
85
+
- Did I use the same sentence structure more than twice in a row?
86
+
- Did I use "not X — it's Y" or "Here's the thing" anywhere?
87
+
- Did I stack three or more historical examples back-to-back?
88
+
- Did I inflate the stakes beyond what the content warrants?
89
+
- Would a human actually write a first draft this way?
90
+
- Does any passage sound like it belongs on a motivational poster?
91
+
92
+
## The One Rule
93
+
94
+
Write like a human: varied, imperfect, specific.
+107
.claude/skills/ai-writing-tropes/references/composition.md
+107
.claude/skills/ai-writing-tropes/references/composition.md
···
1
+
# Composition Tropes
2
+
3
+
## Fractal Summaries
4
+
5
+
"What I'm going to tell you; what I'm telling you; what I just told you" —
6
+
applied at every level of the document. Every subsection gets a summary. Every
7
+
section gets a summary. The document itself gets a summary.
8
+
9
+
Bad:
10
+
11
+
- "In this section, we'll explore... [3000 words later] ...as we've seen in this
12
+
section."
13
+
- "A conclusion that restates every point already made in the previous 3000 words"
14
+
- "And so we return to where we began."
15
+
16
+
Fix: Trust readers to remember what they just read. Summarize only when the
17
+
audience genuinely needs it (executive summaries, abstracts).
18
+
19
+
## The Dead Metaphor
20
+
21
+
Latching onto a single metaphor and beating it into the ground across the entire
22
+
piece. A human writer would introduce a metaphor, use it, then move on. AI will
23
+
repeat the same metaphor 5-10 times.
24
+
25
+
Bad:
26
+
27
+
- "The ecosystem needs ecosystems to build ecosystem value."
28
+
- "Walls and doors used 30+ times in the same article"
29
+
- "Every paragraph finds a way to say 'primitives' again"
30
+
31
+
Fix: Use a metaphor once or twice, then let it go. If you need it again later, it
32
+
should feel like a callback, not a crutch.
33
+
34
+
## Historical Analogy Stacking
35
+
36
+
Rapid-fire listing of historical companies or tech revolutions to build false
37
+
authority. Especially common in technical writing.
38
+
39
+
Bad:
40
+
41
+
- "Apple didn't build Uber. Facebook didn't build Spotify. Stripe didn't build
42
+
Shopify. AWS didn't build Airbnb."
43
+
- "Every major technological shift -- the web, mobile, social, cloud -- followed
44
+
the same pattern."
45
+
- "Take Spotify... Or consider Uber... Airbnb followed a similar path... Shopify
46
+
is another example... Even Discord..."
47
+
48
+
Fix: Pick one example and go deep. A single well-analyzed case is worth more than
49
+
five name-drops.
50
+
51
+
## One-Point Dilution
52
+
53
+
Making a single argument and restating it in 10 different ways across thousands of
54
+
words. The model pads a simple thesis to feel "comprehensive" by rephrasing the
55
+
same idea with different metaphors, examples, and framings. An 800-word argument
56
+
becomes 4000 words of circular repetition.
57
+
58
+
Bad:
59
+
60
+
- "The same point, restated eight ways across 4000 words."
61
+
- "Each section rephrases the thesis with a different metaphor but adds nothing
62
+
new"
63
+
64
+
Fix: State the point. Support it. Move on. If a piece only has one idea, it
65
+
should be short.
66
+
67
+
## Content Duplication
68
+
69
+
Repeating entire sections or paragraphs verbatim within the same piece. Happens
70
+
when the model loses track of what it has already written, especially in longer
71
+
pieces.
72
+
73
+
Bad:
74
+
75
+
- "The same section appeared twice, word-for-word identical."
76
+
- "Paragraph 3 and paragraph 17 are the same sentence reworded"
77
+
78
+
Fix: Re-read what you've written. Cut duplicates.
79
+
80
+
## The Signposted Conclusion
81
+
82
+
Explicitly announcing the conclusion with "In conclusion", "To sum up", or "In
83
+
summary". Competent writing doesn't need to tell you it's concluding. The reader
84
+
can feel it.
85
+
86
+
Bad:
87
+
88
+
- "In conclusion, the future of AI depends on..."
89
+
- "To sum up, we've explored three key themes..."
90
+
- "In summary, the evidence suggests..."
91
+
92
+
Fix: Just write your final thought. Drop the signpost.
93
+
94
+
## "Despite Its Challenges..."
95
+
96
+
The rigid formula where AI acknowledges problems only to immediately dismiss them.
97
+
Always follows the same beat: "Despite its [positive words], [subject] faces
98
+
challenges..." then ends with "Despite these challenges, [optimistic conclusion]."
99
+
100
+
Bad:
101
+
102
+
- "Despite these challenges, the initiative continues to thrive."
103
+
- "Despite its industrial and residential prosperity, Korattur faces challenges
104
+
typical of urban areas."
105
+
106
+
Fix: If challenges matter, discuss them seriously. If they don't, skip them. Don't
107
+
use them as a rhetorical speed bump on the way to optimism.
+42
.claude/skills/ai-writing-tropes/references/formatting.md
+42
.claude/skills/ai-writing-tropes/references/formatting.md
···
1
+
# Formatting Tropes
2
+
3
+
## Em-Dash Addiction
4
+
5
+
Compulsive overuse of em dashes for dramatic pauses, parenthetical asides and
6
+
pivot points. A human writer might use 2-3 per piece naturally; AI will use 20+.
7
+
8
+
Bad:
9
+
10
+
- "The problem -- and this is the part nobody talks about -- is systemic."
11
+
- "The tinkerer spirit didn't die of natural causes -- it was bought out."
12
+
- "Not recklessly, not completely -- but enough -- enough to matter."
13
+
14
+
Fix: Use commas, parentheses, or separate sentences. Reserve em dashes for the
15
+
one or two places where they genuinely earn their keep.
16
+
17
+
## Bold-First Bullets
18
+
19
+
Every bullet point or list item starts with a bolded phrase or sentence. Common in
20
+
Claude and ChatGPT markdown output. Almost nobody formats lists this way when
21
+
writing by hand. A telltale sign of AI-generated documentation and blog posts.
22
+
23
+
Bad:
24
+
25
+
- "**Security**: Environment-based configuration with..."
26
+
- "**Performance**: Lazy loading of expensive resources..."
27
+
28
+
Fix: Write list items as normal sentences. If the list needs headings, use actual
29
+
headings.
30
+
31
+
## Unicode Decoration
32
+
33
+
Use of unicode arrows, smart/curly quotes, and other special characters that can't
34
+
be easily typed on a standard keyboard. Real writers typing in a text editor
35
+
produce straight quotes and `->` or `=>`.
36
+
37
+
Bad:
38
+
39
+
- "Input -> Processing -> Output"
40
+
- "This leads to better outcomes -> which means higher engagement"
41
+
42
+
Fix: Use `->`, `=>`, or just write "leads to". Use straight quotes.
+34
.claude/skills/ai-writing-tropes/references/paragraph-structure.md
+34
.claude/skills/ai-writing-tropes/references/paragraph-structure.md
···
1
+
# Paragraph Structure Tropes
2
+
3
+
## Short Punchy Fragments
4
+
5
+
Excessive use of very short sentences or sentence fragments as standalone
6
+
paragraphs for manufactured emphasis. RLHF training pushes models toward "writing
7
+
for readability" aimed at the lowest common denominator: one thought per sentence,
8
+
no mental state-keeping required. No real person writes first drafts this way.
9
+
10
+
Bad:
11
+
12
+
- "He published this. Openly. In a book. As a priest."
13
+
- "These weren't just products. And the software side matched. Then it
14
+
professionalised. But I adapted."
15
+
- "Platforms do."
16
+
17
+
Fix: Combine related thoughts into real paragraphs. Let sentences build on each
18
+
other. Trust readers to follow compound sentences.
19
+
20
+
## Listicle in a Trench Coat
21
+
22
+
Numbered or labeled points dressed up as continuous prose. The model writes a
23
+
listicle but wraps each point in a paragraph that starts with "The first... The
24
+
second... The third..." to disguise the format.
25
+
26
+
Bad:
27
+
28
+
- "The first wall is the absence of a free, scoped API... The second wall is the
29
+
lack of delegated access... The third wall is the absence of scoped
30
+
permissions..."
31
+
- "The second takeaway is that... The third takeaway is that..."
32
+
33
+
Fix: Either write actual prose that flows between ideas with real transitions, or
34
+
admit it's a list and format it as one. The disguise fools nobody.
+132
.claude/skills/ai-writing-tropes/references/sentence-structure.md
+132
.claude/skills/ai-writing-tropes/references/sentence-structure.md
···
1
+
# Sentence Structure Tropes
2
+
3
+
## Negative Parallelism
4
+
5
+
The "It's not X -- it's Y" pattern, often with an em dash. The single most
6
+
commonly identified AI writing tell. AI uses this to create false profundity by
7
+
framing everything as a surprising reframe. One in a piece can be effective; ten
8
+
in a blog post is an insult to the reader.
9
+
10
+
Includes the causal variant "not because X, but because Y" where every
11
+
explanation is framed as a surprise reveal.
12
+
13
+
Bad:
14
+
15
+
- "It's not bold. It's backwards."
16
+
- "Feeding isn't nutrition. It's dialysis."
17
+
- "Half the bugs you chase aren't in your code. They're in your head."
18
+
19
+
Fix: State the actual point directly. "This is backwards." Done.
20
+
21
+
## "Not X. Not Y. Just Z."
22
+
23
+
The dramatic countdown. AI builds tension by negating two or more things before
24
+
revealing the actual point.
25
+
26
+
Bad:
27
+
28
+
- "Not a bug. Not a feature. A fundamental design flaw."
29
+
- "Not ten. Not fifty. Five hundred and twenty-three lint violations across 67 files."
30
+
- "not recklessly, not completely, but enough"
31
+
32
+
Fix: Lead with the point. "523 lint violations across 67 files."
33
+
34
+
## "The X? A Y."
35
+
36
+
Self-posed rhetorical questions answered immediately. The model asks a question
37
+
nobody was asking, then answers it for dramatic effect.
38
+
39
+
Bad:
40
+
41
+
- "The result? Devastating."
42
+
- "The worst part? Nobody saw it coming."
43
+
- "The scary part? This attack vector is perfect for developers."
44
+
45
+
Fix: Merge into a single statement. "The result was devastating." Or better yet,
46
+
show why it was devastating.
47
+
48
+
## Anaphora Abuse
49
+
50
+
Repeating the same sentence opening multiple times in quick succession.
51
+
52
+
Bad:
53
+
54
+
- "They assume that users will pay... They assume that developers will build...
55
+
They assume that ecosystems will emerge..."
56
+
- "They could expose... They could offer... They could provide... They could
57
+
create..."
58
+
59
+
Fix: Vary your sentence openings. Combine related points. Cut the weak ones.
60
+
61
+
## Tricolon Abuse
62
+
63
+
Overuse of the rule-of-three pattern, often extended to four or five. A single
64
+
tricolon is elegant; three back-to-back tricolons are a pattern recognition
65
+
failure.
66
+
67
+
Bad:
68
+
69
+
- "Products impress people; platforms empower them. Products solve problems;
70
+
platforms create worlds. Products scale linearly; platforms scale
71
+
exponentially."
72
+
- "workflows, decisions, and interactions"
73
+
74
+
Fix: Use two items, or five. Break the rhythm. Not everything needs three beats.
75
+
76
+
## "It's Worth Noting"
77
+
78
+
Filler transitions that signal nothing. AI uses these to introduce new points
79
+
without connecting them to the previous argument.
80
+
81
+
Also includes: "It bears mentioning", "Importantly", "Interestingly", "Notably".
82
+
83
+
Bad:
84
+
85
+
- "It's worth noting that this approach has limitations."
86
+
- "Importantly, we must consider the broader implications."
87
+
- "Interestingly, this pattern repeats across industries."
88
+
89
+
Fix: Delete the filler and state the point. "This approach has limitations."
90
+
91
+
## Superficial Analyses
92
+
93
+
Tacking a present participle ("-ing") phrase onto the end of a sentence to inject
94
+
shallow analysis. The model attaches significance to mundane facts using phrases
95
+
like "highlighting its importance", "reflecting broader trends", or "contributing
96
+
to the development of...".
97
+
98
+
Bad:
99
+
100
+
- "contributing to the region's rich cultural heritage"
101
+
- "underscoring its role as a dynamic hub of activity and culture"
102
+
103
+
Fix: If the analysis matters, give it its own sentence with actual substance. If
104
+
it doesn't matter, cut it.
105
+
106
+
## False Ranges
107
+
108
+
Using "from X to Y" constructions where X and Y aren't on any real scale.
109
+
Legitimate "from X to Y" implies a spectrum with a meaningful middle. AI uses it
110
+
as a fancy way to list two loosely related things.
111
+
112
+
Bad:
113
+
114
+
- "From innovation to implementation to cultural transformation."
115
+
- "From the singularity of the Big Bang to the grand cosmic web."
116
+
117
+
Fix: Just list the things, or pick the one that matters most.
118
+
119
+
## Gerund Fragment Litany
120
+
121
+
After making a claim, AI illustrates it with a stream of verbless gerund
122
+
fragments — standalone sentences with no grammatical subject. The first sentence
123
+
already said everything. The fragments add nothing except word count.
124
+
125
+
Bad:
126
+
127
+
- "Fixing small bugs. Writing straightforward features. Implementing well-defined
128
+
tickets."
129
+
- "Shipping faster. Moving quicker. Delivering more."
130
+
131
+
Fix: If examples help, use a real sentence. "They fix bugs and write
132
+
straightforward features." Or just cut them — the claim already landed.
+138
.claude/skills/ai-writing-tropes/references/tone.md
+138
.claude/skills/ai-writing-tropes/references/tone.md
···
1
+
# Tone Tropes
2
+
3
+
## "Here's the Kicker"
4
+
5
+
False suspense transitions that promise a revelation but deliver a point that did
6
+
not need the buildup. The model manufactures drama before an otherwise
7
+
unremarkable observation.
8
+
9
+
Also includes: "Here's the thing", "Here's where it gets interesting", "Here's
10
+
what most people miss".
11
+
12
+
Bad:
13
+
14
+
- "Here's the kicker."
15
+
- "Here's the thing about AI adoption."
16
+
- "Here's where it gets interesting."
17
+
18
+
Fix: Drop the windup. State the point.
19
+
20
+
## "Think of It As..."
21
+
22
+
The patronizing analogy. AI defaults to teacher mode and assumes the reader needs
23
+
a metaphor to understand anything. Often produces analogies less clear than the
24
+
original concept.
25
+
26
+
Bad:
27
+
28
+
- "Think of it like a highway system for data."
29
+
- "Think of it as a Swiss Army knife for your workflow."
30
+
31
+
Fix: Explain the actual thing. If a metaphor genuinely helps, use it without the
32
+
"think of it as" framing.
33
+
34
+
## "Imagine a World Where..."
35
+
36
+
The classic AI invitation to futurism. Begins with "Imagine" followed by a list
37
+
of wonderful things that will happen if the reader agrees with the premise.
38
+
39
+
Bad:
40
+
41
+
- "Imagine a world where every tool you use has a quiet intelligence behind it..."
42
+
- "In that world, workflows stop being collections of manual steps and start
43
+
becoming orchestrations."
44
+
45
+
Fix: Describe what exists or what you're proposing. Skip the hypothetical
46
+
daydream.
47
+
48
+
## False Vulnerability
49
+
50
+
Simulated self-awareness or honesty that reads as performative. The model pretends
51
+
to break the fourth wall or admit a bias, creating a false sense of authenticity.
52
+
Real vulnerability is specific and uncomfortable; AI vulnerability is polished and
53
+
risk-free.
54
+
55
+
Bad:
56
+
57
+
- "And yes, I'm openly in love with the platform model"
58
+
- "And yes, since we're being honest: I'm looking at you, OpenAI, Google,
59
+
Anthropic, Meta"
60
+
- "This is not a rant; it's a diagnosis"
61
+
62
+
Fix: If you have a bias, show it through your arguments. Don't announce it for
63
+
credibility points.
64
+
65
+
## "The Truth Is Simple"
66
+
67
+
Asserting that something is obvious, clear or simple instead of proving it. If you
68
+
have to tell the reader your point is clear, it probably isn't.
69
+
70
+
Bad:
71
+
72
+
- "The reality is simpler and less flattering"
73
+
- "History is unambiguous on this point"
74
+
- "History is clear, the metrics are clear, the examples are clear"
75
+
76
+
Fix: Present the evidence. Let the reader decide if it's clear.
77
+
78
+
## Grandiose Stakes Inflation
79
+
80
+
Everything is the most important thing ever. AI inflates the stakes of every
81
+
argument to world-historical significance. A blog post about API pricing becomes a
82
+
meditation on the fate of civilization.
83
+
84
+
Bad:
85
+
86
+
- "This will fundamentally reshape how we think about everything."
87
+
- "will define the next era of computing"
88
+
- "something entirely new"
89
+
90
+
Fix: Match your language to the actual stakes. Most things are incremental
91
+
improvements, and that's fine.
92
+
93
+
## "Let's Break This Down"
94
+
95
+
The pedagogical voice that assumes the reader needs hand-holding. AI defaults to a
96
+
teacher-student dynamic even when writing for expert audiences.
97
+
98
+
Also includes: "Let's unpack this", "Let's explore", "Let's dive in".
99
+
100
+
Bad:
101
+
102
+
- "Let's break this down step by step."
103
+
- "Let's unpack what this really means."
104
+
- "Let's explore this idea further."
105
+
106
+
Fix: Just present the analysis. Readers don't need permission to follow along.
107
+
108
+
## Vague Attributions
109
+
110
+
Attributing claims to unnamed authorities instead of being specific. AI invokes
111
+
"experts", "observers", "industry reports" without naming anyone. It also inflates
112
+
quantity — presenting what one person said as a widely held view.
113
+
114
+
Bad:
115
+
116
+
- "Experts argue that this approach has significant drawbacks."
117
+
- "Industry reports suggest that adoption is accelerating."
118
+
- "Observers have cited the initiative as a turning point."
119
+
120
+
Fix: Name the source or drop the attribution. "A 2024 Gartner report found..."
121
+
or just state the claim directly.
122
+
123
+
## Invented Concept Labels
124
+
125
+
AI clusters invented compound labels that sound analytical without being grounded.
126
+
It appends abstract problem-nouns (paradox, trap, creep, divide, vacuum,
127
+
inversion) to domain words and uses them as if they're established terms. They
128
+
function as rhetorical shorthand: name a thing, skip the argument. Multiple such
129
+
labels in the same piece is a strong AI signal.
130
+
131
+
Bad:
132
+
133
+
- "the supervision paradox"
134
+
- "the acceleration trap"
135
+
- "workload creep"
136
+
137
+
Fix: Describe the phenomenon instead of labeling it. If the label isn't
138
+
established in the field, don't pretend it is.
+65
.claude/skills/ai-writing-tropes/references/word-choice.md
+65
.claude/skills/ai-writing-tropes/references/word-choice.md
···
1
+
# Word Choice Tropes
2
+
3
+
## "Quietly" and Other Magic Adverbs
4
+
5
+
Overuse of "quietly" and similar adverbs to convey subtle importance or
6
+
understated power. AI reaches for these to make mundane descriptions feel
7
+
significant.
8
+
9
+
Also includes: "deeply", "fundamentally", "remarkably", "arguably".
10
+
11
+
Bad:
12
+
13
+
- "quietly orchestrating workflows, decisions, and interactions"
14
+
- "the one that quietly suffocates everything else"
15
+
- "a quiet intelligence behind it"
16
+
17
+
Fix: Remove the adverb. If the thing is important, the facts will show it.
18
+
19
+
## "Delve" and Friends
20
+
21
+
"Delve" went from an uncommon English word to appearing in a huge percentage of
22
+
AI-generated text. Part of a family of overused AI vocabulary.
23
+
24
+
Also includes: "certainly", "utilize", "leverage" (as a verb), "robust",
25
+
"streamline", "harness".
26
+
27
+
Bad:
28
+
29
+
- "Let's delve into the details..."
30
+
- "Delving deeper into this topic..."
31
+
- "We certainly need to leverage these robust frameworks..."
32
+
33
+
Fix: Use plain words. "Look at", "use", "strong", "simplify".
34
+
35
+
## "Tapestry" and "Landscape"
36
+
37
+
Ornate or grandiose nouns where simpler words work. "Tapestry" for anything
38
+
interconnected. "Landscape" for any field or domain.
39
+
40
+
Also includes: "paradigm", "synergy", "ecosystem", "framework".
41
+
42
+
Bad:
43
+
44
+
- "The rich tapestry of human experience..."
45
+
- "Navigating the complex landscape of modern AI..."
46
+
- "The ever-evolving landscape of technology..."
47
+
48
+
Fix: Say what you mean. "The field of AI" or just "AI". "How things connect"
49
+
instead of "tapestry".
50
+
51
+
## The "Serves As" Dodge
52
+
53
+
Replacing simple "is" or "are" with pompous alternatives. AI avoids basic copulas
54
+
because its repetition penalty pushes it toward fancier constructions.
55
+
56
+
Also includes: "stands as", "marks", "represents".
57
+
58
+
Bad:
59
+
60
+
- "The building serves as a reminder of the city's heritage."
61
+
- "Gallery 825 serves as LAAA's exhibition space for contemporary art."
62
+
- "The station marks a pivotal moment in the evolution of regional transit."
63
+
64
+
Fix: Use "is". "The building is a reminder of the city's heritage." Shorter,
65
+
clearer, better.