skills.rest
dsgalkar
dsgalkarCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

benchmark-testing

Creates isolated test projects and launches Claude Code sessions to benchmark vercel-plugin skill injection.

Updated Sep 5, 2026
Search Tags:#test automation#claude code#plugin development#wezterm#benchmark testing#skill injection#vercel plugin
One-click install
npx skills add https://github.com/dsgalkar/dnyaneshwar_portfolio --skill benchmark-testing-dsgalkar
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: benchmark-testing
Source: https://github.com/dsgalkar/dnyaneshwar_portfolio/tree/main/.agents/plugins/vercel/.claude/skills/benchmark-testing
Command: npx skills add https://github.com/dsgalkar/dnyaneshwar_portfolio --skill benchmark-testing-dsgalkar

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Manually verifying that a Claude Code plugin injects the right skills across varied scenarios is slow and inconsistent. This Skill automates the creation of isolated benchmark projects and launches parallel Claude Code sessions with crafted prompts to exercise plugin skill injection. ## Core Features & Use Cases - Isolated Test Directories: Creates numbered test project folders under a base directory and installs the vercel-plugin into each via npx add-plugin. - Parallel Session Launching: Spawns WezTerm tabs running Claude Code with correct flags (--cwd, unset CLAUDECODE, --settings .claude/settings.json) and 10-second staggered launches. - Technology-Agnostic Prompt Library: Provides nine ready-made prompts (recipe platform, trivia game, code review bot, and more) mapped to expected injected skills like auth, payments, cron-jobs, and ai-sdk. - Use Case: After modifying the vercel-plugin's skill triggers, run this benchmark to confirm that a prompt like "build a conference ticketing system" still injects the payments, email, and auth skills. ## Quick Start Ask the agent to set up the benchmark test projects and launch Claude Code sessions in WezTerm panes using the provided prompt table.

Frequently Asked Questions about benchmark-testing

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I test that a Claude Code plugin injects the right skills?▼

Create isolated test directories, install the plugin into each with npx add-plugin, then launch Claude Code sessions with crafted prompts describing product features. Compare the injected skills against the expected skills listed in the prompt table.

How do I launch multiple Claude Code sessions in WezTerm?▼

Use wezterm cli spawn with --cwd set to the absolute project path, unset CLAUDECODE to avoid nested-session errors, and pass --settings .claude/settings.json. Wait 10 seconds between launches and always use spawn rather than split-pane.

Why should benchmark prompts avoid naming specific technologies?▼

Naming technologies like Next.js or Stripe bypasses the plugin's inference logic. Prompts should describe the product and features so the plugin itself decides which skills to inject, which is the behavior being benchmarked.

Why does Claude Code fail with a nested-session error in WezTerm?▼

The CLAUDECODE environment variable is inherited from the parent session. Unset it with unset CLAUDECODE before launching the new session inside the spawned pane.

What are the limitations of split-pane for parallel test sessions?▼

WezTerm runs out of usable space after roughly four split panes. The spawn command creates new tabs instead, allowing many more concurrent benchmark sessions without layout constraints.

Related Skills

Explore More Software Engineering Skills→
Adithiya-SAdithiya-S

benchmark-testing

Creates isolated test projects and launches Claude Code sessions to benchmark vercel-plugin skill injection.

Community
Intermediate
Dataslayer-AIDataslayer-AI
18

ds-eval

Evaluate user inputs against skill descriptions from YAML test files.

Official
Intermediate
sourcegraphsourcegraph
32

run-status

Retrieve and summarize active benchmark run statuses within a specified time frame.

Official
Basic
Adithiya-SAdithiya-S

benchmark-agents

Runs interactive Claude Code eval sessions to verify Vercel plugin skill injection and hook behavior.

Community
Advanced