| name | no-tropes |
| description | Detect and eliminate common AI writing tropes from generated text. Use when writing or revising prose, documentation, READMEs, commit messages, PR descriptions, blog posts, comments, or any non-code text output. Automatically applied as a self-revision pass (tropes.fyi). |
No Tropes (AI writing tells)
After drafting any prose, revise it against the tropes catalog in tropes-reference.md.
Core Principle
Write like a human: varied, imperfect, specific. Any single trope used once might be fine. The problem is when multiple tropes stack or a single trope repeats.
Quick Checklist
Before finalizing text, scan for:
- Word choice: "quietly", "delve", "tapestry", "landscape", "serves as", "leverage", "robust"
- Sentence structure: "It's not X — it's Y", "The X? A Y.", tricolons, anaphora, false ranges, "-ing" tacked-on analysis
- Paragraph structure: Short punchy fragments as standalone paragraphs, listicles disguised as prose
- Tone: "Here's the thing", "Think of it as", "Imagine a world", "Let's break this down", invented concept labels, grandiose stakes inflation
- Formatting: Em-dash overuse (2-3 max per piece), bold-first bullets, unicode arrows (→)
- Composition: Fractal summaries, dead metaphors beaten into the ground, historical analogy stacking, one-point dilution, "In conclusion"
Revision Protocol
- Read the full draft
- Flag any trope matches from the reference
- Rewrite flagged passages using plain, direct language
- Prefer simple verbs ("is" over "serves as"), concrete nouns, and varied sentence structure
- Cut filler transitions ("It's worth noting", "Importantly", "Interestingly")
- If you catch yourself writing the same structural pattern 3+ times, break the pattern
