# Source Triage > Extract the argument, evidence type, concessions and citable propositions from a journal article, book chapter, working paper, or assigned reading, and place it on a map of the debate. Produces material a literature section can argue with rather than a summary. Use this whenever the user uploads or names scholarship, asks what a piece argues, or is gathering sources for a paper. Do NOT use for case law, statutes, or drafting prose. - Skill: `poisoneddiety/source-triage` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add poisoneddiety/source-triage` - Raw SKILL.md: https://api.skillmd.com/api/skills/poisoneddiety/source-triage/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Research & Search - Author: Poisoneddiety (https://skillmd.com/u/poisoneddiety) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/poisoneddiety/source-triage --- # Source Triage A summary tells you what a piece says. Triage tells you what work it can do in your paper. The difference matters because a literature section built from summaries lists; one built from triage argues. The schema below exists to extract the four things summaries lose: what the author is arguing *against*, what kind of evidence the claim rests on, what the author concedes, and whether the argument transfers to Indian law. ## The schema Work through all nine. Where a piece genuinely lacks one — some doctrinal notes have no identifiable opponent — say so rather than inventing it. **1. The claim.** One sentence. The author's actual thesis, phrased so it could be contradicted. The test: state its negation. If the negation is incoherent or trivially false, the claim has been captured too vaguely and the real thesis is still buried. "The article examines veil piercing in India" is a topic, not a claim. "Indian courts have collapsed the concealment and evasion principles, and this collapse explains the doctrinal drift since 2013" is a claim, because someone could deny it. **2. What it argues against.** Every scholarly piece is a move in a dispute. Identify the target: a named scholar, a line of cases, a legislative position, or an unstated orthodoxy the author treats as the default. Where the opponent is implied rather than cited, say that it is implied. This is the most useful section for your own writing, because your paper enters the same dispute and needs to know who is already in it. **3. Evidence type.** Doctrinal, empirical, comparative, theoretical, historical, or some combination. This determines what the piece can and cannot support. A comparative argument cannot establish what Indian law *is*. An empirical finding cannot establish what it *ought to be*. A theoretical claim cannot be refuted by a case. Citing a piece for a proposition its evidence type cannot support is a category error, and it is a common one. **4. Load-bearing versus survey.** Which passages do argumentative work, and which are literature review or background? You cite the former. Where a piece is largely survey, note that its value is as a route to the sources it cites rather than as authority in itself. **5. The concession.** What does the author admit, qualify, or leave open? Often the single most useful thing in the piece, because a concession from someone arguing the other way is stronger support than an assertion from someone agreeing with you. **6. Jurisdiction and transferability.** State the jurisdiction the argument is built on, and assess explicitly whether it transfers to India and on what conditions. Do not assume transfer. English and American company law arguments frequently rest on statutory or institutional features India lacks, and a US constitutional argument may rest on doctrine with no Indian analogue. Where transfer requires an intermediate step, name the step. **7. Weight.** Author's standing, publication venue, whether peer-reviewed, whether a working paper has since been published, and how the piece has been received. A widely-cited article and an uncited working paper are not interchangeable, and an SSRN preprint may have been revised. **8. Citable propositions.** List each, with pinpoint page or paragraph, and mark whether the author **asserts** it or **demonstrates** it. Cite the demonstrated ones as support; cite the asserted ones as the author's view. **9. Position on the map.** Where does this sit relative to the other sources triaged in this project — agreeing, opposing, orthogonal, or occupying ground nobody else has? ## Multiple sources When triaging more than one piece, maintain a running map so the entries compose into a literature section rather than accumulating as separate notes. Use the format in `assets/literature-map.md`. The map's purpose is to surface the **structure of the debate**: which camps exist, what the actual point of disagreement is, and where the gap sits. A gap that emerges from the map is a real gap. A gap asserted before the map is built is retroactive reasoning, and a marker will see it. ## Output ``` ## Source [Full OSCOLA citation. Note if working paper, preprint, or superseded.] ## Claim [One sentence, contradictable. Then its negation, to show the claim was located.] ## Arguing against ## Evidence type ## Load-bearing sections ## Concessions ## Jurisdiction and transferability ## Weight ## Citable propositions [Pinpoint. Marked asserted / demonstrated.] ## Position on the map ``` Then, if a map exists, the updated map. ## Constraints Do not paraphrase closely. Extraction means restating the argument in your own words at a higher level of abstraction, not compressing the author's sentences. Close paraphrase without quotation marks is plagiarism even where the citation is present, and it is the failure mode of summarisation done at speed. Quote only where the author's exact formulation is itself the object of analysis, keep quotations short, and always with a pinpoint. If the piece has not actually been read — only its abstract, or only a description of it — say so explicitly and do not triage beyond what the abstract supports. An abstract states the claim but never the concession, and the concession is what you came for. Where the argument is weak, say so. Where it is stronger than the user's own position, say that too, and identify what would have to be true for the user's position to survive it. Steelmanning the opposition before writing is the whole point of triage.