Test Driver

This page is auto-generated from the plugin's README. Do not edit directly — changes will be overwritten on the next sync.

Test Driver

Teaches Claude to proactively drive thorough testing across any project type via gap analysis, convergence loops, and persistent status tracking.

Summary

Claude doesn't systematically think about testing unless asked. This plugin installs an always-on testing mindset that evaluates whether testing is needed at natural breakpoints (feature complete, bug fixed, pre-merge), runs structured gap analysis to find what's missing, and iterates through a convergence loop to generate tests and fix issues until everything passes. Testing posture is tracked in a persistent JSON file so future sessions start with full awareness of the project's test health.

Principles

[P1] Test at Breakpoints, Not Every Edit: Surface testing at natural breakpoints, never after individual edits. Silent change tracking; noisy only when it matters.

[P2] Inline Over Delegated: All analysis and test generation happens in the main context window for maximum code awareness. No agent delegation.

[P3] Converge, Don't Repeat: The convergence loop drives toward all-green with oscillation detection. If the same fixes keep breaking each other, stop and surface the pattern.

[P4] Profile-Driven Stack Knowledge: Framework-specific testing knowledge lives in lightweight, swappable profiles. Adding support for a new stack means adding one file.

Requirements

None beyond Claude Code.

Enhanced by (optional): python-dev, qt-suite, home-assistant-dev plugins.

Installation

/plugin marketplace add L3DigitalNet/Claude-Code-Plugins
/plugin install test-driver@l3digitalnet-plugins

Usage

Automatic (always-on): The testing-mindset skill loads for any implementation task and suggests gap analysis at natural breakpoints.

Manual commands:

/test-driver:analyze   # force a full gap analysis
/test-driver:status    # view current test posture