Persona-based code review prompt generator for VSCode.
Define expert reviewer personas, select one or more, paste a PR URL, and get a structured prompt ready to run in Claude or any AI assistant with GitHub MCP access.
- Define personas — Create reviewer personas like "Security Auditor" or "Performance Expert", each with custom instructions and a checklist.
- Select personas — Pick one or more personas for a given review.
- Enter a PR URL — Provide a GitHub pull request URL.
- Generate — Reviseo builds a multi-persona review prompt.
- Paste into your AI assistant — The prompt instructs the AI to use GitHub MCP tools to fetch the diff, add review comments, and consolidate findings.
The central value is the quality of the generated prompt — each persona reviews independently, then overlapping comments are merged into stronger, consolidated feedback.
- 20 built-in expert personas covering security, architecture, testing, performance, observability, concurrency, clean code, and more.
- Full persona CRUD — create, edit, and delete your own custom personas with a name, instructions, and checklist items.
- Multi-persona composition — select any combination of personas per review.
- Structured prompts — generated prompts include consolidation logic to eliminate duplicate comments.
- GitHub MCP integration — prompts are designed for AI assistants with the GitHub MCP server (
pull_request_read,pull_request_review_write,add_comment_to_pending_review). - Privacy-first — all persona data is stored locally in VSCode global state; nothing is sent to any remote server.
- VSCode theme support — UI adapts to your current light or dark theme.
| Persona | Focus |
|---|---|
| Security Expert | Trust boundaries, injection, secrets, auth |
| Software Architect | Package structure, coupling, resilience patterns |
| Hexagonal Architecture Expert | Domain isolation, ports/adapters, dependency direction |
| Modular Monolith Expert | Vertical slices, module boundaries, data ownership |
| Skeptical Architect | Over-engineering, reinvention, tech fit |
| Solution Design Evaluator | Design alternatives, simplicity vs. complexity |
| Clean Code Expert | Readability, naming, duplication, KISS |
| Testing Expert | Test type selection, mocking, assertion strength |
| Performance Test Expert | Workload realism, warmup, latency percentiles |
| Concurrency Expert | Async correctness, thread safety, context propagation |
| Reliability Engineer | Timeouts, retries, circuit breakers, graceful degradation |
| Backward Compatibility Expert | Breaking changes, API evolution, deprecation |
| Database Interaction Expert | Transactions, N+1 queries, connection pooling |
| Observability Expert | Logging placement, metrics, correlation IDs |
| Azure Pipelines Architect | Stage dependencies, approvals, rollback strategy |
| Spring Batch Expert | Job semantics, restartability, error handling |
| Prompt Engineer | Clarity of intent, role definition, I/O format |
| ...and more |
| Command | Description |
|---|---|
Reviseo: Open Panel |
Open the main Reviseo panel |
Reviseo: Manage Personas |
Open the panel on the Personas tab |
Reviseo: Generate Review Prompt |
Open the panel on the Generate tab |
Access via the Command Palette (Ctrl+Shift+P / Cmd+Shift+P).
To use the generated prompts end-to-end, your AI assistant needs access to the GitHub MCP server. The generated prompt uses these tools:
pull_request_read— fetches the PR diffpull_request_review_write— creates a pending reviewadd_comment_to_pending_review— adds comments to the pending review
Reviseo itself has no external dependencies — it only generates the prompt text.
- Clone the repository:
git clone https://github.com/TaurusOSS/reviseo - Install vsce if you don't have it:
npm install -g @vscode/vsce - Install dependencies:
npm install - Package the extension:
vsce package - Install the generated
.vsixfile in VSCode:- Open the Extensions view (
Ctrl+Shift+X/Cmd+Shift+X) - Click Views and More Actions... (the
...menu) - Select Install from VSIX...
- Pick the
.vsixfile produced in step 4
- Open the Extensions view (
Initial release — persona management, prompt generation, 20 built-in personas.