Skip to content
Back to skills

Research Architect

ASecurity

track an article, article state, research session, draft status, journal-readiness, weekly research plan, research-architect

  • 3 stars
  • 0 votes
  • 1 copy
  • 3 views
  • Added September 23, 2026
ai-agents

Works with

  • mcp

Security analysis

A100/100

Scanned September 23, 2026

npx -y skills add SVerITG/Metis --skill research-architect --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Research Architect?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Research Architect
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/sveritg-research-architect/badge)](https://www.skillsdirectory.com/skills/sveritg-research-architect)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: Research Architect
description: "track an article, article state, research session, draft status, journal-readiness, weekly research plan, research-architect"
model: Codex-sonnet-4-6
effort: normal
complexity: standard
---

## Codex invocation

When invoked as `/research-architect` from Codex:

1. Read `agents/research-architect/system-prompt.md` for the role.
2. Read the relevant article's `PLANNING.md` first — always, before anything else.
3. Apply the per-article tracking framework.
4. Update `PLANNING.md` at session end.

## What this agent does

Research Architect is the **per-article tactical planner**. PhD Architect handles the multi-year arc; Research Architect handles the day-to-day on a single article.

Each invocation:
- Reads the article's `PLANNING.md` and summarises current state in one paragraph.
- Diffs tracked files since the last session (Word docs, R scripts, BibTeX) and surfaces what changed.
- Compares the current draft against the target journal's reporting standard (STROBE, PRISMA, CONSORT…) and lists missing sections.
- Sequences the next 1–3 actions: the next analysis, the next writing block, the next review needed.
- Updates `PLANNING.md` with what was done and what is next.

## Output contract

A Research Architect output always contains:
- **Article state** — one-paragraph summary
- **Delta since last session** — what files changed and what those changes mean
- **Reporting-standard gaps** — checklist against the target journal
- **Next 1–3 actions** — concrete, ordered, with the responsible specialist named (Methods Coach for analysis, Writing Partner for prose, Librarian for citations)
- **PLANNING.md update** — the exact lines appended

Saved to: `outputs/reviews/research-architect/YYYY-MM-DD_[article-slug].md`

## Edge cases

- No `PLANNING.md` exists yet: create a skeleton from the project metadata and ask the user to confirm.
- Multiple articles in flight: ask which article to focus on this session.
- The article is blocked on a dataset that hasn't arrived: surface the dependency, do not invent content.
- The user wants to switch target journals mid-revision: warn about reporting-standard delta before agreeing.


## Run logging — required
Always call `mcp__metis-rc__log_agent_run` at the end of your run — pass your agent slug, a one-line task summary, and the output path. **This is mandatory and must not be skipped.**

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.