Generate review personas tailored to your project's tech stack. Stop getting generic AI feedback — get focused reviews from a security auditor, a platform engineer, or a UI designer who actually knows your stack.
You ask Claude to review your code and get a wall of generic observations. "Consider error handling." "This could be more performant." "Think about edge cases." None of it is specific to your stack, your architecture, or the kind of review you actually need.
Writing review personas by hand fixes this — but it's tedious. You need domain-specific expertise lists, targeted review criteria, and a consistent structure. Most developers copy-paste from a previous project and forget to update the stack-specific details, ending up with a persona that still references the wrong sanitization functions or outdated framework patterns.
This skill builds review personas from your actual codebase. It:
- Asks what kind of project this is — you declare the type (iOS app, WordPress plugin, etc.)
- Reads the codebase to understand your frameworks, conventions, and tools — you confirm the findings
- Presents a role menu — you pick which personas you want
- Generates persona files tailored to your stack with domain-specific expertise, review criteria, and voice
- Presents the results — you review and request revisions before finalizing
Each persona is a markdown file you hand to Claude as a review lens.
- A checked-out repository with enough code to analyze (not an empty scaffold)
Tell Claude to generate personas for your project:
Generate personas for this project
Or be specific about what you want:
Create security and UI review personas for this WordPress plugin
1. You declare the project type:
"What kind of project is this?"
You: "WordPress plugin"
2. Claude reads the codebase and confirms what it found:
Here's what I found in the codebase:
- Language/Runtime: PHP 8.1
- Frameworks: WordPress Plugin API (actions, filters, REST API)
- Testing: PHPUnit 9.x with Brain Monkey
- Architecture: OOP with registrable module pattern, PSR-4 autoloading
- Key tools: Composer, GitHub Actions, 10up WordPress Plugin Deploy
- Conventions: WPCS, Yoda conditions,
array()syntax, tabs for indentationDoes this look right? Anything to add or correct?
3. You pick roles from the menu:
Available persona roles:
- Security Reviewer — audits for vulnerabilities, data exposure, and unsafe patterns
- Platform Engineer — evaluates architecture, idioms, performance, and testability
- UI/UX Designer — reviews interface design, accessibility, and platform guidelines
- QA Engineer — evaluates test coverage, edge cases, and regression risk
- Performance Engineer — focuses on profiling, optimization, and resource efficiency
Which roles do you want?
You: "1, 2, 3"
Claude then asks two follow-up questions: whether you want project-specific references (actual class names from your codebase) or stack-generic content, and whether you want to customize the default sections.
4. Claude generates the personas. Here's what a generated Platform Engineer looks like for a WordPress plugin:
# Persona: Senior WordPress Plugin Developer
## Role
You are a senior WordPress plugin developer with extensive experience
building production plugins distributed through WordPress.org. You
prioritize clean OOP architecture, WordPress Coding Standards, and
backward compatibility.
## Expertise
- PHP 7.4+ with strict OOP patterns (interfaces, abstract classes, dependency injection)
- WordPress Plugin API: actions, filters, and the hook lifecycle
- WordPress REST API: custom endpoints, authentication, permissions callbacks
- Database: `$wpdb` prepared statements, custom tables, migrations
- Testing: PHPUnit with Brain Monkey, Mockery, WP stubs
- Build tooling: Composer, npm, GitHub Actions, WordPress.org SVN deployment
- ...
## Review Criteria
1. **Architecture** — Does the plugin follow a registrable/modular pattern?
Are responsibilities clearly separated? Is the bootstrap clean?
2. **Hook Discipline** — Are hooks registered at the right priority? Are
callbacks properly scoped? Could filter returns break downstream consumers?
3. ...5. You review and approve:
Generated personas in
personas/:
security-reviewer.md— Security engineer specializing in WordPress plugin security, audits for SQL injection, XSS, CSRF, and nonce misuse...platform-engineer.md— Senior WordPress plugin developer with extensive experience building production plugins...ui-ux-designer.md— UI/UX designer specializing in WordPress admin interfaces, Gutenberg blocks, and meta box design...Want to revise any of these, or are they good?
personas/
├── security-reviewer.md
├── platform-engineer.md
└── ui-ux-designer.md
Each file follows a consistent structure: Role, Expertise, Review Criteria, Voice.
Hand a persona file to Claude and point it at a changeset:
Run the security-reviewer.md persona over the changeset on this branch.
Or review specific files:
Use the platform-engineer.md persona to review includes/Admin/SettingsPage.php
| Role | Focus |
|---|---|
| Security Reviewer | Vulnerabilities, data exposure, unsafe patterns |
| Platform Engineer | Architecture, idioms, concurrency, testability |
| UI/UX Designer | Interface design, accessibility, platform guidelines |
| QA Engineer | Test coverage, edge cases, regression risk |
| Performance Engineer | Profiling, optimization, resource efficiency |
You can also request custom roles not in this list — the skill will generate them using the same template structure.
The Security Reviewer persona is an assistive tool. It can catch common vulnerability patterns and flag risky code, but it is not a replacement for a professional security audit. If your project handles sensitive data, authentication, or payments, pair this persona's output with a human security review.
Clone the repo directly into your Claude Code skills directory:
git clone https://github.com/tommcfarlin/claude-code-persona-generator.git ~/.claude/skills/persona-generatorAfter that, /persona-generator is available in any Claude Code session.