Skip to content
Agentmart
Submit a tool

optimize-prompt

Refactor system prompts, tool descriptors, and agent instructions into high-density, contract-driven, XML-structured production specifications.

Skill by mimimaster

IndexedMIT★ 142
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/mimimaster-optimize-prompt

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Refactor system prompts, tool descriptors, and agent instructions into high-density, contract-driven, XML-structured production specifications. Trigger via /optimize-prompt.

Use cases: documents, agent-building

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/optimize-prompt

What's inside

Skill name
optimize-prompt
Description
Refactor system prompts, tool descriptors, and agent instructions into high-density, contract-driven, XML-structured production specifications. Trigger via /optimize-prompt.
License
MIT
Instructions
37 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Optimize Prompt

## Goal
Transform conversational, bloated, or unstructured prompts into high-density, contract-driven production specifications with zero persona fluff, zero strawman arguments, and strict XML/bullet boundaries.

## Done means
- **4-Point Diagnostic Audit** executed on the original prompt:
  1. *Persona & Filler Bloat*: Strip fluff like "You are a world-class...", "Take a deep breath...", "Please be smart/helpful".
  2. *Strawman Lecturing*: Eliminate "why not" preambles, comparisons with non-existent competitors, or defensive justifications; state target architecture affirmatively.
  3. *Attention Dilution*: Convert dense prose paragraphs (wall-of-text) into structured, scannable bullet points.
  4. *Missing Data Envelopes*: Wrap dynamic inputs, tool results, and output schemas in standard `<tag>` boundaries.
- **Production-Grade Rewrite Delivered**:
  - **English Runtime Core**: Ultra-lean, high Token-density English instructions for model execution.
  - **Structured Architecture**: Clear breakdown across `Scope / Trigger`, `Data Contracts`, `Hard Negative Boundaries`, and `Output Format`.
  - **Colocated Chinese Translation**: 1:1 Chinese explanatory translation placed directly underneath for engineering review.
- **1:1 Side-by-Side Comparison**:
  - Direct before vs after diff clearly identifying pruned implementation leaks and token savings.
- **Verification Probe**:
  - Concrete test/probe case or edge scenario demonstrating that the optimized prompt eliminates failure modes (e.g. anti-hallucination, anti-deadlink, double-rendering).

## Stop when
- Prompt is already at peak density (<100 tokens, XML-isolated, bullet-point contracts) — report that no change is necessary.
- Operational intent or domain boundaries are ambiguous — clarify the intended behavior; do not invent domain invariants.

## Constraints
- **Zero Persona / Emotional Fluff**: Models execute operational rules, not roleplay.
- **Zero Negative Strawmen**: Never write "Unlike other bad frameworks..." or "We do this because X is stupid". Only declare what this system *is* and *does*.
- **No Implementation Leaks**: Never expose host internals (e.g. database schema, UI CSS animations, scheduler algorithms) to model-facing tool descriptors.
- **Dual-Language Colocation**: English for production runtime; Chinese for review and documentation.

## Verify
- Re-check the rewritten prompt against:
  1. Token count reduction (typically 30%–60% savings).
  2. Strict XML tag closure and markdown syntax validity.
  3. Presence of explicit negative boundaries (what NOT to do) alongside positive capabilities.

From mimimaster/piwin, MIT.

More from mimimaster/piwin

Compare and alternatives

    Featured in

    Signals

    Stars
    142
    Forks
    6
    Open issues
    0
    Last commit (this folder)
    Aug 31, 2026 (1 month ago)
    Last release
    Oct 8, 2026 (19 days ago)

    From GitHub (fetched Oct 24, 2026).

    Details

    License
    MIT
    First seen
    Oct 24, 2026
    Last checked
    Oct 24, 2026

    More from mimimaster

    Cite this page

    optimize-prompt skill. Agent Mart. https://agentmart.dev/skills/mimimaster-optimize-prompt (last updated 2026-10-24)