Catra's writing voice, extracted from 3,833 tweets and encoded into a single file.
Drop it into any AI. Generate content that sounds like this voice.
- 12 rhetorical moves ranked by frequency — from compressed aphorism (25%) to single-word declarative drop (<1%)
- Word mechanics — recurring vocabulary, register switching, EN/ID code-switching patterns
- Sentence mechanics — opening types, closing devices, length distribution, punctuation rules
- 15 anti-patterns — what this voice never does, as important as what it does
- 7 rewrite pairs — generic phrasing transformed into the voice, for AI calibration
- Reference tweets — real examples pulled verbatim from the corpus, organized by pattern
- 3,833 tweets analyzed
- 3 batch analyses conducted independently, then cross-validated
- Engagement-filtered — corpus limited to tweets at or above median engagement (likes + retweets)
- Spans topics: on-chain creation, creator sovereignty, minimalism, fitness, indie culture, electronic music, web3
[paste catra.md]
Write a post about [topic].
The constraints do the work.
catra.md— the voice profilePROCESS.md— how it was built
The file isn't the moat. The person is.