nevaberry-plugins
Nevaberry · Nevaberry/nevaberry-plugins
Updates AI coding agents with current technology knowledge, syntax, and APIs to overcome LLM knowledge cutoff limitations, ensuring accurate and modern code generation.
Knowledge Patch is a fleet of SKILL.md files that fill the gap between an LLM's knowledge cutoff and today.
Preview
What you can do with it
- Ensure AI-generated code uses the latest syntax and APIs for modern frameworks like Next.js or Dioxus.
- Prevent AI agents from suggesting deprecated methods or outdated best practices.
- Provide AI with up-to-date information on specific technologies like Kubernetes or RHEL.
Safety check
No risky patterns found in the files we scanned.
Automated scan of the repo files. Always review code you run.
From the README
nevaberry-plugins
Knowledge Patch is a fleet of SKILL.md files that fills current technology knowledge gaps without mixing target-model extraction streams.
Nevaberry mines technologies such as Kubernetes, Bun, Next.js, RHEL, Dioxus, and Rust, then compresses each target’s own extraction stream into topic-organized skills. Claude and Codex receive parallel trees with identical frontmatter and target-specific bodies.
Why: Claude or Codex loaded with the knowledge-patch plugin uses the newest syntax and API on the first try, instead of confidently generating last year's deprecated API.
https://nevaberry.com/en/knowledge-patch
Install
Claude Code:
/plugin marketplace add Nevaberry/nevaberry-plugins
/plugin install knowledge-patch@nevaberry
/knowledge-patch:knowledge-patch-setup
Codex: codex plugin marketplace add Nevaberry/nevaberry-plugins, install knowledge-patch from /plugins, start a new session, then run $knowledge-patch-setup.
The plugin is fully offline, works in both CLIs, and needs no credentials. See plugins/knowledge-patch/README.md for skills, activation, hooks, and versioning.
Knowledge patches are derived from the respective projects' official release notes and documentation. This project is not affiliated with or endorsed by those projects.