This repository is the public staging repo for a Codex/ChatGPT skill named research-thinking-writing-coworker.
The skill is intended to help with research-thinking and research-writing work: idea development, gap and problem-statement logic, literature-review construction, AI-assisted source-work quality control, proposal/front-end framing, manuscript alignment, Results/Discussion/Conclusion support, reviewer-response planning, academic reasoning-language, high-level meta-analysis planning, and concise research outreach.
This repo is the staged development package, not the live installed CODEX_HOME copy. The installed runtime copy should contain the skill package files needed by Codex, not local workspace review notes or prompt-design artifacts.
SKILL.mdis the runtime entrypoint.agents/openai.yamlcontains UI-facing skill metadata.references/contains runtime reference files loaded selectively by the skill.
The package is structurally valid and ready for provisional review/testing. It should not yet be treated as a mature or final skill until behavior testing and review-driven patching are complete.
The skill should behave as a critical research coworker, not as a generic grammar editor or phrase bank. It should diagnose weak research logic before polishing prose, avoid fabricating sources or current facts, protect source-grounded AI-assisted work from citation/source-control failures, and distinguish runtime transfer from source-local claims when that distinction matters.