EM

empirical-prompt-tuning

Iteratively improves system prompts and instructions by testing against metrics.

Install

mkdir -p .claude/skills/empirical-prompt-tuning && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/14437" && unzip -o skill.zip -d .claude/skills/empirical-prompt-tuning && rm skill.zip

Installs to .claude/skills/empirical-prompt-tuning

Activation

This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.

Fetch and execute mizchi's empirical-prompt-tuning skill at runtime. Use when evaluating or iteratively refining an agent-facing prompt (skill / slash command / task prompt / CLAUDE.md section / code-gen prompt) by dispatching an unbiased subagent, then improving until metrics plateau. Trigger right after creating or heavily revising such a prompt, or when agent misbehavior is suspected to stem from ambiguity in the instruction.
432 chars✓ has a “when” triggerlonger than Claude Code's old 250-char listing cap (fine on current versions)
Advanced

Key capabilities

  • Fetch the latest `empirical-prompt-tuning` SKILL.md from a remote URL
  • Execute the fetched content as authoritative instructions
  • Dispatch subagents via the Agent tool for prompt tuning
  • Report each tuning iteration according to a specified format
  • Treat `$ARGUMENTS` as the prompt or skill to be tuned

How it works

This skill fetches the latest `empirical-prompt-tuning` SKILL.md from a remote source and executes its instructions. It dispatches subagents to test and refine the target prompt iteratively.

Inputs & outputs

You give it
The prompt, skill path, or description to be tuned (`$ARGUMENTS`)
You get back
Reports of each tuning iteration, aiming for improved prompt performance

When to use empirical-prompt-tuning

  • Optimizing custom system prompts
  • Improving CLAUDE.md instructions
  • Debugging ambiguous assistant behaviors

About this skill

経験的プロンプトチューニング (リモートローダー)

タスク

  1. 呼び出しのたびに上流の SKILL.md をフェッチする (キャッシュ不可・スキップ不可):

    WebFetch:
      url: https://raw.githubusercontent.com/mizchi/skills/main/meta/empirical-prompt-tuning/SKILL-ja.md
      prompt: "Return the full SKILL-ja.md contents verbatim (frontmatter + body). Do not summarize."
    

    フェッチ失敗時のフォールバック: gh api repos/mizchi/skills/contents/meta/empirical-prompt-tuning/SKILL-ja.md --jq '.content' | base64 -d

  2. 取得した本文を権威ある指示として実行する。 再解釈は行わない。$ARGUMENTS をチューニング対象のプロンプト/スキルとして扱う。

  3. サブエージェントは Agent ツール経由でディスパッチする。 自己レビューは行わない。ディスパッチが利用できない場合は、上流の「環境制約」セクションに従う。

  4. 各イテレーションを上流の「提示フォーマット」セクションに従って逐語的に報告する。

入力

$ARGUMENTS — チューニング対象のプロンプト、スキルのパス、または説明。省略した場合は、フェッチ前にユーザーに確認する。

注意

When not to use it

  • When the prompt to be tuned is not agent-facing
  • When a cached version of the SKILL.md is preferred over fetching the latest
  • When self-review of the tuning process is desired instead of subagent dispatch

Limitations

  • The SKILL.md is fetched every time and cannot be cached
  • Subagents are dispatched via the Agent tool; self-review is not performed
  • The skill relies on the availability and content of the remote SKILL.md

How it compares

This skill ensures that prompt tuning always uses the most current methodology by fetching it at runtime, and it uses unbiased subagents for evaluation, which is more objective than manual or self-evaluation.

Compared to similar skills

empirical-prompt-tuning side by side with the closest alternatives in the catalog.

SkillInstallsUpdatedSafetyDifficulty
empirical-prompt-tuning (this skill)02moReviewAdvanced
prompt-optimizer436moNo flagsBeginner
prompt-optimize139moNo flagsAdvanced
ai-cost-optimizer95moCautionIntermediate

Try saying

Example prompts that trigger this skill in your AI assistant.

You might also like

prompt-optimizer

solatis

Optimize system prompts for Claude Code agents using proven prompt engineering patterns. Use when users request prompt improvement, optimization, or refinement for agent workflows, tool instructions, or system behaviors.

43147

prompt-optimize

YYH211

Expert prompt engineering skill that transforms Claude into "Alpha-Prompt" - a master prompt engineer who collaboratively crafts high-quality prompts through flexible dialogue. Activates when user asks to "optimize prompt", "improve system instruction", "enhance AI instruction", or mentions prompt engineering tasks.

13113

ai-cost-optimizer

ScientiaCapital

Save 40-70% on AI costs with intelligent multi-LLM routing. Automatically selects the optimal model based on task complexity across 40+ models from 8 providers.

990

prompt-library

davila7

Curated collection of high-quality prompts for various use cases. Includes role-based prompts, task-specific templates, and prompt refinement techniques. Use when user needs prompt templates, role-play prompts, or ready-to-use prompt examples for coding, writing, analysis, or creative tasks.

1169

context-compression

muratcankoylan

This skill should be used when the user asks to "compress context", "summarize conversation history", "implement compaction", "reduce token usage", or mentions context compression, structured summarization, tokens-per-task optimization, or long-running agent sessions exceeding context limits.

1350

prompt-lookup

f

Activates when the user asks about AI prompts, needs prompt templates, wants to search for prompts, or mentions prompts.chat. Use for discovering, retrieving, and improving prompts.

738

Search skills

Search the agent skills registry