prompt-token-efficiency

Rewrites prompts to reduce token count while preserving intent and clarity.

5.1k|530|Updated Aug 12, 2025
One-click install
npx skills add https://github.com/github/gh-aw --skill prompt-token-efficiency
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: prompt-token-efficiency
Source: https://github.com/github/gh-aw/tree/main/.github/skills/prompt-token-efficiency
Command: npx skills add https://github.com/github/gh-aw --skill prompt-token-efficiency

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Prompts written for humans often contain filler, repetition, and vague wording that waste tokens and introduce ambiguity when consumed by LLMs. This Skill compresses prompts into minimal, unambiguous instructions without losing required constraints or output quality.

Core Features & Use Cases

  • Token Reduction: Removes pleasantries, narrative framing, and filler words while keeping task-critical information.
  • Ambiguity Elimination: Resolves undefined nouns, unclear pronouns, and vague qualifiers like "appropriate" or "some" into measurable criteria.
  • LLM-Optimized Formatting: Converts prose into imperative statements, atomic bullets, and explicit output format requirements.
  • Use Case: Before sending a long instruction block to an agentic workflow, run it through this Skill to cut token usage and make success criteria testable.

Quick Start

Rewrite this prompt to use fewer tokens while keeping all constraints and making the output format explicit.

Frequently Asked Questions about prompt-token-efficiency

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

FAQPage Schema
How do I reduce token count in a prompt without losing meaning?

Remove pleasantries, repetition, and narrative framing, then keep only task-critical facts. Replace long qualifiers with concrete limits and end with explicit success criteria so the compressed prompt preserves all required constraints.

How to write prompts optimized for LLM execution?

Use imperative statements, prefer bullets over long prose, and keep each instruction atomic. State the objective in one short sentence first, put required context before optional context, and specify the output format unambiguously.

What makes a prompt ambiguous for an LLM?

Ambiguity comes from undefined nouns, pronouns without clear antecedents, vague words like "appropriate" or "some", and missing scope limits. Verify that time ranges, quantities, and success criteria are explicit and testable before finalizing.

When should I not compress a prompt aggressively?

Avoid aggressive compression when examples are needed to prevent failure, since removing them can degrade output quality. Keep one minimal example if the task is prone to misinterpretation, and never drop required constraints to save tokens.

Does shorter prompt text always improve LLM output quality?

No. Shorter prompts help only when clarity and constraints are preserved. The goal is minimal tokens with maximal clarity, so conflicting instructions or missing context must be resolved rather than deleted.