addyosmani/agent-skills: Production-grade engineering skills for AI coding agents.
Production-grade engineering skills for AI coding agents. - addyosmani/agent-skills
Open Source Prompt Library
Summary
A collection of concise, human-written modular prompts designed to optimize the performance and output style of AI agents. These 'microskills' aim to reduce verbosity and provide specific functional capabilities like controlled prose length and session debriefing.
Highlights
- Focuses on handwritten, non-verbose instructions to prevent context pollution in LLMs.
- Compatible with skills.sh and any agent capable of reading SKILL.md files.
- Easy installation via command line using 'npx skills add staltz/microuskills'.
- Includes specialized commands such as '/peep' for controlling word counts and '/debrief' for summarizing complex decision paths.
auto-generated
Context
Audience
Developers and AI Engineers
DomainArtificial Intelligence
FormatGitHub Repository
Accessopen source
Source Overview
A repository of concise, human-written prompts for AI agents, installable via npx and compatible with skills.sh and any SKILL.md reader.
Contains skills like PEEP (word-capped plain English), DEBRIEF (agent decision summaries after long sessions), and WHERE (code location shortcuts). Designed to reduce context pollution from verbose, agent-generated instruction files.
Topics
Related
Prompt EngineeringAI AgentsLarge Language ModelsLLM Orchestrationskills.sh
