AI Agent Hub
Back to skills
Prompt Optimizer icon

Prompt Optimizer

AI Agent Updated 2026.08.30

Paste the following prompt into your AI chat to install this skill:

Please follow https://skillhub.cn/install/skillhub.md to install @user_dca6a55a/prompt-plus.

About this skill

Problem

Prompt writing often stalls on unclear intent, missing constraints, and non-reusable experience. prompt-plus is for engineers and agent developers who repeatedly create, diagnose, and refine Prompt artifacts, turning vague requirements into reviewable, saved, searchable prompt assets.

Workflow

The skill provides three modes:
- Create mode: runs without parameters, moving through requirement clarification, scenario deep-diving, structure draft, generation review, and persistence.
- Optimize mode: takes an existing Prompt, restores intent, decomposes structure, scores quality, matches patterns, then outputs before/after changes and reasoning.
- Learn mode: uses --learn to analyze strong examples, extract role setting, step-by-step guidance, output control, and other techniques, then writes patterns into knowledge/patterns/ indexes.

The key principle is understand first: each round asks at most 4 questions, challenges vague phrases such as “better,” and preserves original intent while explaining what each change fixes.

Boundaries

It fits Chinese-language, maintenance-oriented team workflows that rely on AskUserQuestion and a Markdown knowledge base. It is less suitable for one-off simple instructions, zero-interaction batch pipelines, or environments where target model, input/output format, and business constraints are unavailable; result quality still depends on the supplemental information supplied.

Use Cases

  • Turn vague writing requirements into reusable prompts with role, constraints, and output format.
  • Diagnose ambiguity, context gaps, and formatting constraints in existing prompts, then show before/after.
  • Extract roles, step-by-step guidance, and output control from strong prompts into searchable patterns.
  • Complete input/output and boundary constraints before drafting customer-support, code-review, or summarization agents.

Best For

  • Engineers maintaining multi-turn agent prompts who need to distill scattered experience into reusable patterns.
  • Product or ops owners turning business requirements into prompts who need role, constraint, and output-format clarity.
  • Technical reviewers of existing prompts who need to identify ambiguity and context gaps with reasoned fixes.
  • ML engineers studying strong prompts who need to extract techniques and add searchable knowledge entries.