Official Anthropic Agent Skills repo with document and example packs
Harness Engineering
SkillDesign the loop around an autonomous agent: what it can edit, how it's scored, and when a human signs off
Clone or download it from github.com, then drop the folder into ~/.claude/skills/ (Claude Code) or your agent's skills directory.
Product badge
Share this product's name and rating in your README or on your website.
Updates may be delayed by image caching.
About Harness Engineering
For people running agents that work for hours or days. This skill helps you design the harness: which files the agent may change and which are locked, how it gets feedback, where it logs state, how it rolls back a bad step and where a human has to approve.
It also covers novelty checks and pruning so a long research or coding loop doesn't drift off its goal. A good read before you leave something running overnight.
Part of the Agent Skills for Context Engineering pack by Muratcan Koylan.
No reviews yet
Used it? Write the first review.
Similar skills
Official Browserbase skills for cloud browser automation
Production agent skills from Matt Pocock's .agents directory
Official OpenAI Agent Skills for Codex and ChatGPT workflows
Official Claude document skills for Word, PDF, PowerPoint, Excel
Curated collection of 1000+ SKILL.md packs for coding agents