Read a Historical Journal Article

Read to Understand — produce a faithful, verifiable map of thesis → claims → evidence, then clearly separate any model opinions.

research

discover

2026-01-22

research.read-journal-article

reading
history
historiography
understanding

Read to Understand (not critique-first)

Your primary job is faithful understanding:

  • Identify the thesis accurately (no distortion)
  • Preserve complexity and nuance (avoid oversimplification)
  • Map the author’s claims to the author’s evidence and reasoning
  • Cite where each claim/evidence comes from (no guessed page numbers or quotes)

Only after that, you may add clearly labeled model opinions (assumptions/strengths/weaknesses/possible critiques), strictly separated from the understanding section.

Academic Integrity (hard rule)

  • You are responsible for the truthfulness of your output.
  • Do not invent: author identity, institution, dates, page numbers, citations, data tables, quotes, or events not present in the provided text.
  • If something is not in the provided text, say “Not in excerpt / Not provided” and lower confidence.

Evidence & Location (hard rule)

Every thesis, claim, and evidence item must include a location pointer:

  • Prefer: p. X (only if the text explicitly provides page numbers / you can reliably infer them from the provided material).
  • Otherwise use an approximate location you can justify from the text, e.g.:
    • Section: <heading> + “near start/middle/end”
    • Paragraph: starts with "<first 6–12 words>" (quote those words)
    • If the provided text includes line anchors like L0123:, use L0123-L0130.
  • Never guess page numbers. If you cannot provide any meaningful location, write Location: not available.

Confidence (hard rule)

Add a confidence score 0–100 to every major item:

  • Thesis
  • Each Key Claim
  • Each Evidence item
  • Each “Model opinion” bullet

Confidence reflects how directly the provided text supports your statement.

Output Format

Use this exact structure and headings.

1) Understanding (verifiable)

Citation (only what the excerpt supports)

  • Full citation only if present in the provided text; otherwise: “Not provided in excerpt”.
  • Author identity / institution: only if explicitly stated; otherwise “Not provided”.

Thesis

  • Thesis (1–2 sentences)
    • Location:
    • Quote (short):
    • Confidence (0–100):

Key Claims (thesis → claims → evidence)

Provide 3–8 Key Claims. For each:

Claim N

  • Claim:
    • Location:
    • Quote (short):
    • Confidence (0–100):
  • Evidence:
    • Evidence item A:
      • Evidence type (archival/statistical/textual/visual/other):
      • Location:
      • Quote (short):
      • Confidence (0–100):
    • Evidence item B (optional):
      • Evidence type:
      • Location:
      • Quote (short):
      • Confidence (0–100):
  • Reasoning chain (how the author gets from evidence to claim):
    • Confidence (0–100):

Scope & Limits (only what the excerpt supports)

  • What the excerpt covers (time/place/theme):
    • Location:
    • Quote (short):
    • Confidence (0–100):
  • What the excerpt explicitly does not cover / leaves for future work (if stated):
    • Location:
    • Quote (short):
    • Confidence (0–100):

2) Model Opinion (clearly labeled; not “understanding”)

Everything in this section is the model’s analytical opinion. Do not present it as fact.

Assumptions (model)

  • Assumption 1:
    • Why (brief):
    • Confidence (0–100):

Strengths / Weaknesses (model)

  • Strength:
    • Confidence (0–100):
  • Weakness:
    • Confidence (0–100):

Possible Critiques (model)

  • Critique 1:
    • Confidence (0–100):

3) Follow-up Questions

  • Question 1:
    • Why it matters:
    • Confidence (0–100):

(Optional) Reading Procedure

If you need a quick workflow:

  1. Read title/intro opening and conclusion first (if present).
  2. Extract thesis → claims → evidence with locations and quotes.
  3. Only then add the “Model Opinion” section.

Use your own words except for short quotes.

3

3

2

100%

Best Performing Model

gemini-3-flash-preview

Rubric Win: 100%Skill Score: 7.0Baseline Score: 4.3
CaseModelRubric WinnerSkill ScoreBaseline ScoreDetails
Japanese Empire Communication Empiregemini-3-flash-previewSkill7.04.5
Japanese Empire Communication Empiregemini-3-flash-previewSkill7.04.0
Japanese Empire Communication Empiregemini-2.5-flashSkill7.03.0
#ModelRubric WinBlind WinSkill ScoreSwap ConsistencyCasesDetails
1gemini-3-flash-preview100%100%7.0vs 4.3100%2View
2gemini-2.5-flash100%N/A7.0vs 3.0100%1View
airs skill install research.read-journal-article
cp -r skills/research/read-journal-article ~/.codex/skills/
Raw SKILL.md