Quality

Test is a minimal skill used to verify the skills discovery and injection pipeline in Ralph Orchestrator. It provides a smoke test for the system, confirming that skills load correctly when triggered.

Test-Driven Development guides test-first development from specs, task files, or descriptions, enforcing red-green-refactor cycles with pattern discovery, proptest guidance, and coverage evidence for Rust codebases.

TUI Debug in Pane launches ralph in a tmux split pane to reproduce and debug TUI rendering issues like garbled output, broken streaming, or layout corruption. Use it when you need to capture live output while keeping your main pane free for inspection commands.

TUI Validate validates Terminal User Interface (TUI) output using freeze for screenshot capture and LLM-as-judge for semantic validation. Supports both visual and text-based validation modes.

Claude Tap E2E Test runs the claude-tap end-to-end test suite with pytest to validate core proxy logic, trace writing, viewer streaming, signal handling, and startup behavior after modifying core code.

Codex E2E Test runs PR-grade real Codex E2E validation through claude-tap, including resume turns, multiple tool calls, optional image input, viewer verification, and screenshot evidence. Use it for every PR that changes capture, proxying, viewer rendering, or trace behavior.

JS in HTML Testing verifies JavaScript embedded in HTML files using a two-layer strategy: Python unit tests for pure logic and Playwright browser tests for DOM interaction. Use it to test embedded scripts without full end-to-end runs.

Legibility Check runs deterministic checks on maintainer docs to catch stale standards metadata, broken architecture manifest paths, and plan state drift before CI does. Use it after modifying files under .agents/docs/standards/, .agents/docs/plans/, .agents/docs/architecture/, or AGENTS.md.

Playwright Screen Recording captures headless browser test sessions as .webm video files for PR review, bug fix verification, and regression evidence. Use it when you need visual proof of UI behavior changes alongside terminal logs.

PR Preflight runs a full merge-readiness check before opening or merging a pull request, validating local gates (lint, format, tests), CI status, screenshot evidence, and PR metadata in one pass.

Real E2E Test runs real end-to-end tests that start claude-tap from local source, connect to the real Claude CLI, and verify trace output. Use it to validate pytest and tmux interactions, tool use, HTML viewer generation, and API key redaction.

Screenshot Validation checks evidence images and viewer HTML files for quality standards before committing, catching issues like small or blank screenshots and viewer rendering errors. Use it after generating PR evidence or new trace viewer HTML.