agentic-beta-testers
AI-powered browser testing against any URL using personality-driven agents -- speedrunners, chaos gremlins, skeptical executives -- each sur
Open source Open in the app JSON README (API)
About
AI-powered browser testing against any URL using personality-driven agents -- speedrunners, chaos gremlins, skeptical executives -- each surfacing different categories of usability and reliability issues.
Details
- Kind
- Plugins
- Topic
- Developer tools
- Publisher
- rjspence3
- Origin
- marketplace
- Category
- ferramentas
- Stars
- 1
- Last push
- 2026-03-26T18:32:48Z
- Repository state
- ativo
- Added
- 2026-08-30 01:48:58
- Updated
- 2026-08-30 01:48:58
- Origin id
rjspence3/claude-beta-tester/agentic-beta-testers
README
# claude-beta-tester > A Claude Code plugin that runs AI-powered browser testing against any URL using personality-driven agents — speedrunners, chaos gremlins, skeptical executives, and more. ## Plugin Installation (Claude Plugin Directory) Once listed in the Claude plugin directory: ```bash claude plugin install claude-beta-tester@claude-community ``` Commands will be available as `/claude-beta-tester:beta-test`. ## Manual Install ```bash # Clone into your Claude config directory git clone https://github.com/rjspence3/claude-beta-tester ~/.claude/plugins/claude-beta-tester # Copy command cp ~/.claude/plugins/claude-beta-tester/.claude/commands/beta-test.md ~/.claude/commands/ ``` Command is available immediately in Claude Code as `/beta-test`. ## What it does `/beta-test` runs the `ai-beta-test` CLI against a target URL with one or more personality-driven agents. Each agent has a different mindset and surfaces different categories of issues. ### Available personalities | Name | Mindset | What it surfaces | |------|---------|-----------------| | `speedrunner` | *"I don't have time for this."* | Missing loading states, unclear primary actions, broken fast-paths | | `chaos_gremlin` | *"What happens if I do this?"* | Input validation gaps, error handling issues, state corruption | | `skeptical_exec` | *"I don't trust this tool yet."* | Cognitive load, trust breaks, unexpected transitions | | `methodical_newcomer` | *"I am new here."* | Missing guidance, confusing labels, dead ends | | `overloaded_manager` | *"I have 30 seconds."* | Poor hierarchy, information overload, slow landing | | `calm_operator` | *"Slow and steady."* | Missing feedback, inconsistency, unclear state | | `adhd_founder` | *"Oh look, a shiny feature!"* | Poor resumption, brittle state, punitive flows | | `privacy_paranoid` | *"Why do you need to know that?"* | Trust breaks, missing provenance, dark patterns | | `feature_explorer` | *"Show me everything."* | Undiscoverable features, navigation gaps | | `interactive_explorer` | *"I'll use every single one."* | Incomplete flows, missing feedback | | `technical_exploit` | *"I don't play by the rules."* | Security hints, exposed storage, validation bypasses | | `hybrid_auditor` | *"I read the blueprints."* | Security risks, logic errors, API failures | | `agentic_interface_tester` | *"AI interface UX researcher."* | Mode selection errors, decision transparency | **Default set:** `speedrunner`, `chaos_gremlin`, `skeptical_exec` ## Usage ``` /beta-test ``` Claude will ask for: 1. Target URL 2. Goal (one sentence — e.g., "Complete the onboarding flow") 3. Personalities to run (or say `default`) Or provide them directly: ``` /beta-test https://myapp.com ``` ## Requirements Requires `ai-beta-test` CLI and Node.js 18+: ```bash # From the repo (development) pip install -e ~/Development/ai-beta-tester # Or once published pip install ai-beta-tester # Node.js (required for Playwright MCP server) brew install node ``` --- Built by [Rob Spence](https://nomouthlabs.com). MIT License.