litreviewlisted

# /litreview — Multi-source literature search and synthesis The goal is to give the user a high-signal, ranked, deduped paper list with enough takeaway per item that they can decide which to read, which to cite, and which to skip — all in one pass. The output is the **input** to `/draft` (when writing a lit review) and to `/cite` (when batch-adding the best hits to Zotero). ## Modes - **Default** (no flag): single 1–5 relevance score per paper — exactly as documented in step 3 below. This is the canonical mode. - **`--four-axis`** (opt-in): replace the single score with four 1–5 sub-scores — **Novelty** (vs. the existing lit on this topic), **Credibility** (venue tier + author rep + sample / data quality), **Relevance** (fit to the specific project context), **Actionability** (cite-as-prior / replicate-method / pivot-target / superseded). **Composite** = arithmetic mean of the four, rounded to one decimal. The breakdown is rendered only when **composite >= 4** (else show only the composite, to keep output tight). All downstream thresholds — ranking, in-Zotero check, the `/cite` handoff — use the composite as a drop-in replacement for the single score, so the **composite >= 4** cutoff matches the default **score >= 4** cutoff exactly. Default behavior is unchanged unless `--four-axis` appears in the invocation. ## When to invoke The user invokes `/litreview <query>` with a topic, paper, or research question. Common shapes: - `/litreview discrimination in online labor mar