Openai Responses

This wraps OpenAI's Responses API, the stateful conversation interface they launched in March 2025 that preserves reasoning state across turns instead of discarding it like Chat Completions does. The main advantage is continuity: the model keeps its internal reasoning notebook between messages, which apparently gives a 5% bump on TAUBench. You'd reach for this when building multi-turn agent workflows where context and reasoning chains matter more than one-off completions. Fair warning though, it failed the Gen Agent Trust Hub audit while passing Socket and getting a warning from Snyk, so you'll want to review the security posture before dropping it into production. Over 350 installs suggests people are experimenting with it despite the mixed audit results.

Install to Claude Code

npx -y skills add jezweb/claude-skills --skill openai-responses --agent claude-code