plugin-testing

Benchmark harness for A/B testing Claude Code plugins against OOLONG long-context reasoning tasks. Compare truncation vs RLM-RS recursive chunking strategies. Features Claude Code hooks integration, SQLite persistence, and comprehensive scoring aligned with the OOLONG paper methodology.

This repository hosts the pre-built web output (flutter build web) of a Flutter project used to test and demo plugin behavior in the browser. It is not the source code — only the deployable artifacts (main.dart.js, CanvasKit, assets, and the service worker) live here.

Trustworthy, automated testing for REAPER plugins and extensions - point it at a plugin and get a real pass/fail, no test code required. (macOS today; Windows/Linux WIP)

Vendor-neutral research umbrella for measuring AI plugin, agent, and MCP server quality across CLI runtimes (Claude Code, Gemini CLI, Copilot CLI, Codex CLI).

🚀 Automate the evaluation of Claude Code plugin components to ensure accurate triggering of skills, agents, commands, and hooks.

Binary-criteria evaluation harness for Claude skills with planned extension to plugins, agents, and MCP servers. Score every change yes/no across 7 layers — package integrity, trigger quality, functional quality, regression protection, baseline value, model variance, rollout safety. Never gradients.

Lifecycle conformance testing for DeepSeek Harness plugins

Testing harness for AI agent skills and plugins — contract-driven YAML specs with mocked environments

Independent GitHub Action that proves DeepSeek Harness plugins install, compose, boot, and answer HTTP in a clean profile.

Isolated install-and-smoke test drives for DeepSeek Harness plugins: installs a repo or npm package into a throwaway DSH_HOME profile, verifies the bundle patch layer and boot logs, records a structured pass/fail result matrix (JSON/Markdown) for scoring pipelines, and quarantines every temp directory it owns

Performed end-to-end manual testing and source code analysis on the easy.jobs WordPress Plugin (v2.7.1), identifying 8 bugs across 22 test cases covering authentication, job management, candidate management, security, and edge cases.

Testing framework for Codify plugins with complete lifecycle testing, IPC communication, and cross-platform support

Deterministic, answer-keyed Obsidian vaults for testing plugins, plus faithful hand-authored starter vaults for PKM methods and roles. MIT + CC0.