Persistent project-scoped knowledge base for Claude Code and Codex: quick and deep research depths, contrarian pass, source-independence rules measured on DeepResearch Bench II. Tested end to end on Haiku agents and a local 35B under noob-cli. Installs via npx skills add, /plugin install, codex marketplace, or git clone.
benchmark research memory knowledge-base codex progressive-disclosure openai-codex subagent agent-skills deep-research claude-code claude-skill codex-plugin npx-skills deepresearch-bench
-
Updated
Jul 23, 2026 - Shell