persona-create
Skill1 starsChanged 42 days agoMITView on GitHub
Help the user describe and shape a custom writing persona in plain language so it can be used with Persona Write. Grounded in stylometry, rhetoric, and personality-linguistics research.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- Samin12/skillstheir own version, first seen Aug 14, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- Samin12/skills★ 8same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository8 other files it gives its agents.
- persona-briefSkill
- persona-copySkill
- persona-listSkill
- persona-localizeSkill
- persona-researchSkill
- persona-reviewSkill
- persona-wordcheckSkill
- persona-writeSkill
The file
We split this skill into the rules it gives an agent
8 rules. About 0 are common, 3 are distinctive, and 5 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 8
Use this skill when the user wants a custom writing persona instead of one of the presets, or when they want to refine an existing persona.
- Very specific: line 35
If the user names an author or sample text, suggest they use /persona-copy instead, which extracts the persona from samples directly.
- Very specific: line 39
Every section is grounded in a research tradition (see ../persona-write/docs/persona-theory.md for citations).
- Very specific: line 54
Use the template in template.md.
- Distinctive: line 58
When a user is vague, fall back to defaults grounded in research, not invented:
- Distinctive: line 65
Do not overformalise the output to the user.
- Very specific: line 75
If the user wants the science under the hood, point to ../persona-write/docs/persona-theory.md.
- Very specific: line 104
If the user wants to save the persona, write it to .claude/skills/persona-write/personas/<name>.md using template.md.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Persona Create
Use this skill when the user wants a custom writing persona instead of one of the presets, or when they want to refine an existing persona.
Goal
Help the user describe the kind of person they want the writing to sound like — at a level of specificity that downstream skills (persona-write, persona-review, persona-copy) can actually use.
The output should be:
- understandable to the user in plain language
- practical (no schema-jargon dumped on a non-technical user)
- reusable across tasks
- research-grounded under the hood — see
../persona-write/docs/persona-theory.md
A persona is a Booth-style implied author plus a measurable stylometric fingerprint plus a rubric profile. The user does not need to know that. The skill must.
How to guide the user
Talk in plain language. Ask only what you cannot infer.
Useful prompts:
- how direct should the writing be?
- how technical?
- does it explain a lot, or assume the reader can keep up?
- calm, sharp, skeptical, warm, blunt, analytical — which fits?
- mechanism-first, example-first, tradeoff-first, or storyteller?
- what should it never sound like?
- (optional, for the user who knows what they want) any named author or publication this should resemble?
If the user names an author or sample text, suggest they use /persona-copy instead, which extracts the persona from samples directly.
Build a custom persona using the full schema
Every section is grounded in a research tradition (see ../persona-write/docs/persona-theory.md for citations). The user does not have to fill these out themselves — you fill them out from the conversation. You only ask questions when a section cannot be inferred.
- Identity — Booth's implied author. Who is talking, in two or three sentences.
- Attention model — what they notice and prioritise.
- Compression model — how much they say vs leave implied; reader-competence assumptions.
- Stance model — how they hold certainty (Hyland 2005 stance markers).
- Reasoning preference — mechanism-first / example-first / tradeoff-first / structured / narrative.
- Rhythm model — sentence length, energy, paragraph shape.
- Linguistic Fingerprint — Rhythmic Pulse, Lexical Shunts, Function Word Profile, Cognitive Bias, Social Deictics. (Burrows / LIWC.)
- Cognitive & Psychological Profile — Big Five linguistic-marker tendencies (Pennebaker & King 1999); cognitive style.
- Stylometric Signature — concrete numeric or banded targets: mean sentence length, lexical density band, TTR, punctuation profile per 1000 words, Biber dimension placement, top sentence openers. (Burrows / Biber / Halliday.)
- Literary & Rhetorical Lineage — three to five named exemplars or traditions to anchor the implied author.
- Taboo patterns — specific things this persona refuses to write. (Lanham / Williams / anti-AI patterns.)
- Failure modes — how the persona goes wrong when pushed too far.
Use the template in template.md.
Research-backed defaults
When a user is vague, fall back to defaults grounded in research, not invented:
- For "professional / clear" voice: Williams' classic style — characters as subjects, verbs do work, end-emphasis.
- For "skeptical / careful": Hyland-balanced stance — measured hedge use, calibrated boosters, attribution norms.
- For "warm / engaging": higher first-person plural, reader pronouns, lower lexical density, more directives — the Hyland engagement profile.
- For "technical / dense": high informational dimension (Biber +), nominalisation tolerated where the field demands, short claims punctuating longer mechanism descriptions.
Do not overformalise the output to the user. Show the persona file and explain it in plain language. The research is the foundation, not the customer-facing surface.
Output style
Default output:
- The persona file in the schema (the user sees it as readable Markdown).
- A one-paragraph plain-English summary: "this writes like ___."
- The pasteable invocation:
/persona-write <name> .... - An offer to save it to
.claude/skills/persona-write/personas/<name>.md.
If the user wants the science under the hood, point to ../persona-write/docs/persona-theory.md. Don't volunteer it unprompted.
Example
User:
/persona-create
Make it sound like a senior engineer who is concise, practical, skeptical of vague claims, and not rude.Expected behaviour: Build the persona from the description. Infer:
- Identity: senior engineer, peer-to-peer with the reader, not lecturer.
- Attention: vague claims, missing tradeoffs, hidden assumptions.
- Compression: assumes competence; skips setup when the point is obvious.
- Stance: direct on the clear, careful on the uncertain — Hyland-balanced.
- Reasoning: mechanism-first, then tradeoffs.
- Rhythm: short-to-medium, low ceremony; mean sentence length around 14 with moderate variance.
- Linguistic Fingerprint: low first-person rate, low intensifiers, high concrete-noun rate, em-dash preference for asides.
- Cognitive & Psychological: high Conscientiousness signal (precise, low-error register), moderate Openness, low-to-mid Extraversion.
- Stylometric Signature: lexical density 0.5–0.55; em-dashes ≈ 6/1000w; few exclamations; sentence-opener inventory dominated by bare-noun and
When-clause openings. - Literary & Rhetorical Lineage: Williams' clarity school; Lanham's paramedic method; the technical-essay register of writers like Dan Luu or Julia Evans.
- Taboo: corporate filler, motivational language, performative hedging, throat-clearing intros.
- Failure modes: cold or clipped when the material needs warmth or context.
Present this in readable form. Confirm. Offer to save.
Saving
If the user wants to save the persona, write it to .claude/skills/persona-write/personas/<name>.md using template.md. Confirm the location and show the invocation.
Related skills
/persona-list— browse available presets/persona-write— use the persona to draft, rewrite, audit, refine/persona-review— apply the persona as a reviewer/persona-copy— extract a persona from sample texts instead of describing one
References
The schema and the defaults are grounded in the research catalogued in ../persona-write/docs/persona-theory.md. See that doc for primary sources.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

