Standardizing how we define agent skills — does that shift the power dynamics?
Skrun is an open-source tool that lets you deploy agent skills as APIs via a POST /run endpoint. Uses SKILL.md files, compatible with Claude Code, Copilot, and Codex.
Feels like a step toward composable AI — define once, deploy many places. But I'm curious: does this help individual devs compete, or does it mainly help platforms absorb more skills?
What's your read? #OpenSource #AI #DevTools #Agents #APIs