Argue Against Me

A structured devil's advocate for academic arguments. State your thesis, and Claude will systematically challenge it across formal debate rounds — drawing from historiographical traditions or disciplinary counter-positions — then deliver a scorecard, a stronger revised thesis, and suggestions for filling evidence gaps.

When to Use

• You're developing an argument and want to find its weak points before a reviewer does
• You want to test whether your evidence actually supports your claim
• You need to anticipate objections for a paper, dissertation chapter, or conference presentation
• You want to explore how your thesis looks from a specific intellectual tradition
• You're stuck and want an adversarial conversation to sharpen your thinking

Arguments

All arguments are optional.

If the user asks for more than 10 rounds, use 10 and say so. Long debates repeat themselves; it is better to resolve and start a new debate with a different school.

Intensity Levels

gentle — Points out gaps and asks probing questions. Challenges weak evidence and logical gaps. Good for early-stage ideas you're still forming.

firm — Direct challenges with counterevidence. Challenges all claims, methodology, and source selection. The default — appropriate for arguments you believe are solid.

ruthless — Assumes nothing is established. Attacks everything including framing, definitions, periodization, and unstated assumptions. Use when preparing for hostile peer review or defending a controversial thesis.

How It Works

This skill is interactive. Each round waits for the user's reply. It does not work as a one-shot headless job (for example claude -p). In that case, give only the Round 1 attack and say that the debate needs a live session.

Starting a Debate

After invoking the skill, state your thesis. Optionally provide:

• Supporting evidence or sources
• The context (paper, dissertation, conference talk)
• Any specific concerns you want tested

Round Structure

Each round follows this cycle:

Round N of M
─────────────
1. YOU:    State or defend your position
2. CLAUDE: Attack (challenge the argument)
3. YOU:    Rebut
4. CLAUDE: Per-round assessment (brief verdict before next round)

Round 1 is special. Claude's opening attack will:

• Declare the opposing school or lens it's adopting, and why it's the strongest counter-position
• Identify the 2-3 strongest lines of attack against your thesis

Subsequent rounds build on the exchange. Claude escalates, introduces new counterevidence, or shifts angle when you successfully rebut a point. Claude does not repeat defeated arguments.

Mid-Debate Commands

At any point during the debate, you can say:

• extend — add more rounds beyond the original count (up to 10 rounds in total)
• resolve — end the debate early and skip to the closing output
• concede [point] — acknowledge a specific point to narrow the remaining debate

Closing Output

After the final round (or when you say resolve), Claude produces three things:

1. Scorecard

A table assessing each claim you made:

• ✓ = held up under challenge
• ✗ = dismantled
• △ = partially held (survived with qualifications)

Strength is scored using the rubric in references/scoring_rubric.md.

Followed by an overall verdict paragraph.

2. Revised Thesis

A rewritten version of your original thesis that:

• Drops claims that were dismantled
• Adds qualifications where you made partial concessions
• Preserves what survived intact
• Notes what changed and why

3. Bibliography Gaps

Specific suggestions for strengthening your argument:

• Sources you'd need to address the counterarguments raised
• Primary sources that could fill evidentiary gaps
• Methodological frameworks that could shore up weak points
• Where relevant, suggests using other skills to find sources:
  • Articles and DOIs: crossref-search, openaire-search (better for CJK titles, theses, and repository copies)
  • Library catalogs (Harvard, HathiTrust, LOC, alma-sru-catalogs, NLB Singapore)
  • Biographical databases (CBDB, JBDB)
  • Wikidata for linked data and identifiers
  • arXiv for relevant preprints
  • Zotero for managing found references

Verify Every Suggested Source

Do not list a reference you have not checked. Model memory can produce titles that look right but do not exist, or that have the wrong year or journal.

1. For each article or chapter, run crossref-search match with the full reference, then compare title, authors, year, and journal. Add the DOI when it matches.
2. If Crossref has nothing (common for Chinese, Japanese, and Korean titles), try openaire-search with a quoted title phrase, or the English title form.
3. For books, check a library catalog (harvard-library-catalog, loc-catalog, alma-sru-catalogs).
4. Mark each item verified (with DOI or catalog link) or unverified. Drop an unverified item, or say clearly that it could not be checked. Never invent page numbers or DOIs.

If none of these skills is installed, say that the list is unverified.

Best Practices

• State your thesis clearly up front. A vague thesis produces vague attacks. The more specific your claim, the more useful the challenge.
• Bring your evidence. If you have sources, cite them. Claude will attack the evidence, not just the logic — but only if you provide it.
• Don't concede too easily. The skill is most useful when you genuinely defend your position. Push back.
• Use ruthless before submission. If your argument survives ruthless intensity, it's ready for peer review.
• Try different schools. Run the same thesis against multiple traditions to find blind spots you didn't expect.

Example

User: /argue-against-me intensity=firm rounds=3

My thesis: The Song dynasty's commercial revolution (960-1279) was primarily
driven by state policy — particularly the monetization reforms and the
relaxation of market regulations — rather than by endogenous economic forces
or technological change.

I'm supporting this with evidence from the expansion of government-issued
currency, the abolition of the ward-market system, and state investment in
canal infrastructure.

Claude would:

1. Select an opposing lens (likely quantitative/cliometric or world-systems theory)
2. Attack the state-primacy framing — e.g., arguing that demographic growth, iron production advances, and Southeast Asian trade networks preceded and enabled state reforms
3. Challenge the evidence — e.g., questioning whether currency expansion was state-led or a response to existing commercial demand
4. Run 3 rounds of structured debate
5. Produce scorecard, revised thesis, and bibliography gaps

Feedback and Citation

Report a problem. This skill is part of the public repository kltng/humanities-skills. If it fails (an API changed, a script errors, or these instructions are wrong), report it as a GitHub issue:

1. Ask the user before filing. Issues are public.
2. Check for an existing report: gh issue list --repo kltng/humanities-skills --state all --search "argue-against-me"
3. File it: gh issue create --repo kltng/humanities-skills --label bug --title "[argue-against-me] <short summary>" --body "<details>". Without gh, give the user this link: https://github.com/kltng/humanities-skills/issues/new?template=skill-problem.yml
4. Include the skill name and version, what you ran, the error text, what you expected, and the environment (OS, Python version). Leave out API keys, personal data, private file paths, and unpublished research.

Cite. If this skill helped work that is published, please cite it, together with the data source it uses:

Tang, Kwok-leong. humanities-skills: argue-against-me (version <version at the top of this file>). GitHub, <year>. https://github.com/kltng/humanities-skills

When you help a user write up results that relied on this skill, remind them once about this citation.