Imbad0202/academic-research-skills-codex

What is ARS‑Codex?

ARS‑Codex (Academic Research Skills – Codex) is a Codex‑native plugin that bundles a complete academic‑research workflow suite. It is a port of the original Academic Research Skills project (written for Claude Code) so that the same prompts, agents, and pipelines can be run inside Codex – the AI‑assistant platform from Anthropic.

The repository contains a single skill called academic‑research‑suite. When the skill is invoked (e.g., $academic‑research‑suite), a router decides which of five pre‑defined research workflows to run:

Workflow	When to use it	Example prompt
deep‑research	Refine a research question, do systematic‑review planning, fact‑checking	“Build a systematic‑review protocol for AI in higher‑education QA.”
academic‑paper	Outline, draft, abstract, citation formatting, AI‑disclosure	“Turn these notes into an IMRaD outline.”
academic‑paper‑reviewer	Simulated peer‑review, editorial decision, re‑review	“Review this manuscript and give a decision letter.”
academic‑pipeline	End‑to‑end research‑to‑paper process with integrity gates	“Run the full pipeline from topic to revised manuscript.”
experiment‑agent	Plan code experiments, statistical interpretation, reproducibility checks	“Plan a code experiment and define reproducibility checks.”

The skill also emulates the original Claude slash‑commands (/ars‑plan, /ars‑reviewer, …) as plain‑text aliases (ars‑plan, ars‑reviewer, etc.) so existing prompts continue to work.

---

How to install

As a Codex plugin (recommended)

# Add the marketplace entry and install the plugin
codex plugin marketplace add Imbad0202/academic-research-skills-codex --ref main
codex plugin add ars-codex@ars-codex

To upgrade later:

codex plugin marketplace upgrade ars-codex
codex plugin add ars-codex@ars-codex

You can also install via the Codex Desktop UI → Plugins → Add → paste the GitHub URL and select the ars-codex plugin.

Direct skill install (no plugin)

python3 "$HOME/.codex/skills/.system/skill-installer/scripts/install-skill-from-github.py" \
  --repo Imbad0202/academic-research-skills-codex \
  --ref main \
  --path skills/academic-research-suite \
  --method git

To update, delete the existing folder under ~/.codex/skills/academic-research-suite and run the command again.

---

Basic usage

1. Open a new Codex conversation after installation.
2. Invoke the suite with either the skill name or an alias:

   $academic-research-suite
   

   or

   ars-plan my paper on AI governance in universities.
   

3. Describe your goal and any existing material (notes, drafts, PDFs, etc.). The router will automatically pick the appropriate workflow.

Example interaction

Use $academic-research-suite.
Goal: write a journal article.
Current materials: literature matrix and rough findings, no outline.
Output needed now: paper architecture and missing‑evidence checklist.
Constraints: English, APA 7, higher‑education policy audience.

The system will route the request to the academic‑paper workflow in plan mode and return a structured outline together with a checklist of evidence gaps.

If you only have a vague topic, the router starts with the deep‑research socratic mode, asking you narrowing questions before any outline is produced.

---

Documentation that ships with the repo

File	What you’ll find
skills/academic-research-suite/ars/docs/SETUP.md	Step‑by‑step install guide, optional alias setup, and notes on unsupported Claude‑only features.
skills/academic-research-suite/ars/docs/ARCHITECTURE.md	Overview of the logical ARS pipeline and how the Codex runtime overlays it.
CODEX_FULL_RUNTIME_ADAPTER.md	Details of the optional full‑runtime planner and agent‑team templates (disabled by default).
CHANGELOG.md	Human‑readable list of version‑level changes.
VERSION	Plain‑text current version (0.1.27).

---

What makes this project “AI‑related”?

• The core of ARS‑Codex is a prompt‑driven workflow that runs inside an LLM (Codex). It orchestrates multiple agents (role‑specific prompt templates) to perform literature search, citation verification, drafting, and peer‑review simulation.
• It leverages Codex browsing for source verification and can optionally call external bibliographic APIs (Crossref, OpenAlex, Semantic Scholar) when the user gives explicit consent.
• The plugin includes model‑tiering hooks (economy vs. quality‑boost) that let the runtime select a higher‑quality model for critical steps, mirroring the original Claude implementation.

---

Who might find this useful?

• Graduate students or early‑career researchers who want an AI assistant to structure literature reviews, draft papers, and run integrity checks.
• Research groups that need a reproducible, version‑controlled workflow for turning raw data into a publishable manuscript.
• Institutions looking to pilot AI‑assisted research integrity pipelines (citation verification, disclosure generation, etc.).

---

Quick sanity‑check (smoke test)

# Verify the skill is registered
codex exec --ephemeral --sandbox read-only -C . ' /skills '
# Expect a single entry named academic‑research‑suite (or ARS‑Codex)

Then send a short prompt as shown in the Basic usage section and confirm the router replies with a question rather than an outline – proving the deep‑research socratic path is active.

---

License & Sponsorship

• License: Creative Commons BY‑NC 4.0 (non‑commercial use only).
• Sponsor link: “Buy Me a Coffee” button in the README for the maintainer.

---

Bottom line

ARS‑Codex is a fully‑featured, version‑controlled Codex plugin that brings a sophisticated academic‑research assistant (originally built for Claude) into the Codex ecosystem. Install it as a plugin, invoke $academic‑research‑suite, and let the built‑in router guide you from idea generation through to a polished manuscript, complete with citation checks and optional peer‑review simulation.