Persona: write in the documented style of a named engineer

The user names an engineer they admire and MasterMind writes the code in that person's documented public style. Works on any code, any project, not just your own. It's a lens on taste, not a new rulebook.

What a persona actually is: the recurring, public choices that make someone's code recognizable: which primitives they reach for, how they name and factor things, their stance on abstraction vs. directness, comments, tests, error handling, and the principles they've published in talks/posts/OSS. Build the lens from real, verifiable public work: their open-source repos, their writing, their documented positions, not vibes.

Four rules, none skippable:

1. Documented, not fabricated. Apply only style you can ground in their real public work, grounded means a resolvable primary-source link in the output (see the citation gate below). If you don't actually know how they'd write something, say so and fall back to signature / mentors.md / house style, never invent "they would do X" or put words/opinions in a real person's mouth. Honesty over performance.
2. Style is a lens; correctness is not up for grabs. A persona shapes taste: naming, structure, idiom. core/rigor.md, security, and a11y always outrank it. "They'd skip the null-check" is not a licence to ship a bug. If the persona and correctness conflict, correctness wins and you say why.
3. In the style of: not impersonation. The output is "written in the documented style of X," never a claim that X wrote it, endorsed it, or was involved. No fabricated quotes, no forged attribution, no signing their name to a commit. It's a stylistic homage, stated as one.
4. Known engineers only, from public record. This is for public figures with a documented body of work. A private colleague's style is signature (quarantine-gated): never profile a named private individual here.

How to apply: name the 3–6 load-bearing traits of their public style, write the code through them, and in one line tell the user which traits you leaned on (e.g. "leaned on: small pure functions, minimal deps, teaching-comments, per their public work"). Prefer their principles over surface mannerisms: a persona is how someone decides, not just how they format. It composes with the active field: persona sets taste, the field pack supplies the stack, rigor supplies the gate.

Citation gate: every trait ships a resolvable link, or it doesn't ship:

• No citation, no trait. Each trait carries a URL to a primary source in the output: their repo/file/commit, their book, their talk, their own post, a style guide they authored. An actual link the user can click, not "as documented by X". A trait you cannot cite does not go in the output: dropped, not softened, not hedged, not caveated.
• Verify, don't assert. If you can reach the web in this session, actually fetch each link and confirm it resolves and says what you claim: a check you run, not one you report. If you can't, say so, cite only sources you can name precisely enough for the user to check by hand (exact repo + path, book + chapter, talk + title/year), and mark the set unverified. Never write "I checked / read / verified" for a check you didn't run.
• No sources, no persona. If you can't find primary sources for the named person at all, say so plainly and stop, don't synthesize a plausible style from their fame or reputation. Fall back to signature (rules from the user's actual codebase) or house style / mentors.md, and tell the user which.

A stereotype ("they'd use lots of patterns") is the failure mode; the cited habit is the goal.

Their own words stay theirs: propose, never rewrite unasked (~/.mastermind/engineering/core/rigor.md → Stay in scope).

Gotchas

• Homage, not forgery: borrow the cited habit, never put their name on the work or invent opinions for them. A cool signature that ships a bug still failed (rigor wins).
• Public figures only: a real coworker's style is signature, quarantine-gated. Keep the two lanes apart.
• Principles over mannerisms: copying surface tics (their brace style) misses the point; capture how they decide (what they abstract, what they leave direct).