Skip to content

Claude Cowork Plugin

🧪Prompt Engineer

Build evals, A/B test prompts, audit skills, and benchmark LLM outputs at production quality

4
commands
2
skills

The Prompt Engineer plugin helps you ship prompts and skills that survive production. Every command works standalone with your input — describe the prompt, skill, or eval scenario and get a structured draft ready for your review.

Use /eval-rubric to build a custom evaluation rubric for a skill or prompt covering accuracy, tone, compliance, and latency. Use /test-batch to run a prompt against 10–100 synthetic test cases and produce a coverage and quality report. Use /skill-version to version-control a skill, A/B test two variants, and output a statistical-significance verdict. Use /skill-audit to run static analysis on a SKILL.md — checking frontmatter, edge cases, prompt-injection surface, and length budget.

The plugin's skills activate automatically when relevant: Prompt Optimization Loop covers test case design, edge cases, adversarial inputs, and iteration based on eval results, plus Skill Benchmarking measures latency, accuracy, token cost, and token-budget compliance across skill variants.

Commands

Click a command to copy it. To run it in the Claude app, type it and pick it from the menu that appears; pasting it sends plain text instead of running it.

/eval-rubric

Build a custom evaluation rubric for a skill or prompt covering accuracy, tone, compliance, and latency

/test-batch

Run a prompt against 10–100 synthetic test cases and produce a coverage and quality report

/skill-version

Version-control a skill, A/B test two variants, and output a statistical-significance verdict

/skill-audit

Static analysis of SKILL.md — checks frontmatter, edge cases, prompt-injection surface, and length budget

Skills

Skills activate automatically when Claude Cowork detects you're working on relevant tasks — no slash command needed.

Prompt Optimization Loop

Test case design, edge cases, adversarial inputs, and iteration based on eval results

Skill Benchmarking

Latency, accuracy, token cost, and token-budget compliance measurement across skill variants

Usage examples

/eval-rubric Build a rubric for a customer-support triage skill covering accuracy, empathy, no-refund-promises compliance, and latency

Press Enter to copy this example prompt.

/test-batch Generate 50 synthetic cases for /draft-cold-email mixing happy path, edge cases, and prompt-injection adversarial inputs

Press Enter to copy this example prompt.

/skill-version A/B test variant A (prod) vs variant B (draft) on a 40-case eval set with paired bootstrap; output verdict and effect size

Press Enter to copy this example prompt.

/skill-audit Audit this SKILL.md for frontmatter completeness, prompt-injection surface, edge cases, and 2000-token budget compliance

Press Enter to copy this example prompt.

Install this plugin

Claude desktop app (Cowork)

  1. Click Customize in the left sidebar, open the Plugins tab, then Add → Add marketplace → Add from a repository.
  2. Paste alexclowe/awesome-claude-cowork-plugins and click Sync.
  3. Under Discover, click Add next to Prompt Engineer.

Requires a paid Claude plan. Anthropic is merging Cowork into the main Claude app; this path is the same either way.

Claude Code (terminal)

/plugin marketplace add alexclowe/awesome-claude-cowork-plugins/plugin install prompt-engineer@awesome-claude-cowork-plugins

Run inside Claude Code. Add the marketplace once, then install any plugin by name.

All content generated by this plugin is for drafting and informational purposes only. The prompt engineer is responsible for reviewing, verifying, and customizing all outputs before professional use. This plugin does not constitute professional advice.