Convex Agent Skills drops for AI-native backends

Convex is doubling down on the "AI-native backend" thesis by making their platform internals fully discoverable to coding agents like Cursor and Claude Code.

• –Standardizes the `SKILL.md` format, making complex backend architecture patterns portable across different AI IDEs and CLI agents.
• –Eliminates common LLM hallucinations regarding Convex's unique reactive mutations and queries by providing explicit, deterministic guidance.
• –Focuses on the "developer experience for agents," a growing meta-category where tools are optimized for AI consumption rather than just human readability.
• –Integration with `@convex-dev/eslint-plugin` ensures that agent-generated code isn't just functional, but meets strict linting and performance standards.
• –Reflects a shift in the ecosystem where frameworks are now shipping "agent-ready" configuration alongside their core libraries.