proto-personas
Personas built from the research you have, every claim traced to a source or marked as an assumption.
A persona called “Maria, 34” with hobbies and a stock photo gives a team nothing to test. These are hypotheses about segments, and every line says whether it comes from a source or is a guess.
- 2026
- Claude Code
- Web search, Python 3
- MIT
View on GitHub (opens in a new tab)
git clone https://github.com/VladPolyakov-art/proto-personas ~/.claude/skills/proto-personasWhat it writes
Fixed file names, so the other skills in the kit find them without asking. Each file also gets a row in research/INDEX.md.
- research/personas.md
- The personas, the primary one with its justification, a validation plan and assumption statistics.
- research/personas/evidence-ledger.md
- Every fact and gap the personas rest on.
How it works
- Roles, not charactersNo stock names, photos or hobbies. Each persona has a job story, behaviour, frustrations and choice triggers.
- Every item feeds an insightInsights read “Because [fact] → we need [decision]”. An item that changes no design decision is deleted.
- Assumption share by scriptA small Python script counts how much of each section is assumption. A high share is reported as the finding.
What the agent does
- ReconnaissanceReads the brief, notes and sibling files. With no material, asks 3 to 5 questions.
- Evidence ledgerFacts with their sources, gaps with what was searched.
- Persona setUsually three segments, one primary, plus a stress persona card.
- ProfilesJob to be done, behaviour, frustrations, choice triggers, 3 to 5 insights.
- Primary and planThe primary persona with its justification, and what to check with real users.
- MeasureCounts the assumption share and self-checks against a scored checklist.
Works with the kit
None of these is required. With an empty research/ folder it works from what you give it, and when a sibling’s file is there, it reuses it instead of redoing the work.
Reads
- competitive-researchNamed competitors, their pricing and review pains.
- deep-researchSegments, behaviours and market context.
- ux-auditUsability pains in current solutions.
- simulated-interviewSharpens the validation plan, never counted as evidence.
Read by
- simulated-interviewTakes a persona card, by its slug or the Primary persona heading.
- feature-researchWho needs the feature, and how well that is backed.
- ux-auditThe persona for the walkthrough.
- deep-researchSegments, and assumptions to confirm or weaken.
- competitive-researchWho the user is, in their own words.
Part of the Product research kit: six skills, one research/ folder and one set of evidence tags.