Show HN: NPM/uv for Claude Code – install skills from GitHub with one command
github.comGot tired of manually copying files to .claude/ and managing skills across projects. So I built agr (agent-resources) — basically open source npm/uv but for Claude Code.
The gist:
uvx agr add anthropics/skills/skill-creator
(Or pip install agr if you want it permanently)
Link: https://github.com/kasperjunge/agent-resources
Format is <username>/<repo>/<resource-name> - works for skills, commands, or subagents.
If the github repo is named "agent-resources" you can skip <repo> in the handle.
Why I built this: - One command installs from GitHub — no more copying files to .claude/ - Basically treat .claude/ like .venv/ — let agr manage it - agr.toml tracks your dependencies so teammates can run agr sync and get the same setup - Bundle skills + commands + subagents together into installable packages
Some community examples:
agr add maragudk/skills/bluesky # Post to Bluesky agr add dsjacobsen/golang-pro # Go dev toolkit agr add madsnorgaard/drupal-expert # Drupal expertise
Working on: - Rules support - Cursor + auto-sync between tools
Let me know what you think! This is a fascinating pattern—treating 'agent skills' as composable dependencies rather than monolithic prompts.
I'm curious about the execution model: How are you handling the security implications of an agent pulling and executing arbitrary skill code? Is there an inherent sandboxing layer for these skills, or do they inherit the full privileges of the host agent?
In my experience with agentic tools, managing the 'permissions scope' of 3rd party capabilities is the hardest part of moving from demo to production.