# X Article Review

> Review an article, title, thumbnail, and supporting files for factual support, reader usefulness, and consistency. Use for a publication-readiness check or a focused critique of a draft.

- Skill: `codejunkie99/x-article-review` (Agent Skill, multi-file: 3 files)
- Install (CLI): `npx skillmds@latest add codejunkie99/x-article-review`
- Raw SKILL.md: https://api.skillmd.com/api/skills/codejunkie99/x-article-review/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: codejunkie99 (https://skillmd.com/u/codejunkie99)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/codejunkie99/x-article-review

---


# X Article Review

Review the actual draft and requested assets. Use supplied evidence and verify changing claims through available primary sources where needed. Keep the review proportional to the task; a caption does not require an entire production audit.

## Ask concrete questions

- Does the article deliver the title's promise, and does the thumbnail explain the same subject?
- Can a reader identify the value in the opening? Does the TLDR give a useful shortcut?
- Which claims are demonstrated, firsthand, attributed, estimated, or proposed? Are numbers attached to the correct person and time period?
- Where would the intended reader get stuck? Are prerequisites, steps, exact locations, prompts, and expected results sufficient for the promised guide?
- Does the visual remain readable at its intended size? Check labels, connectors, crop, hierarchy, and the specified ratio.
- Do the files and links included in the handoff exist, open, and support the claims made about them?

A style checker cannot prove originality or correctness. A model rating its own draft 10/10 is not a reliable stopping condition. A correlation in post analytics is not proof that a hook or thumbnail caused the result.

## Return useful findings

Lead with material issues. For each, identify the passage or asset, why it matters, the evidence or missing evidence, and a concrete correction. Distinguish a required correction from a subjective preference. If a requested check could not be performed, state exactly what remains unverified.

When asked to fix, make the authorized edits and check the changed areas. Stop when material issues are resolved; do not loop indefinitely or keep rewriting accepted content. If no material issues remain, say so briefly instead of inventing findings.

Publication is a separate action. Do not post, send, or push solely because review was requested; if publication was already explicitly authorized, do not ask for the same permission again.

See [review format](references/review-format.md) for a compact evidence-led report.