Research

Use when `evidence-review` needs an operational protocol before screening and extraction.

Post-merge paper-voice gate: inspect the reader-facing draft for planner-talk and axis-label artifacts, then route fixes to the actual source.

Audit/regression checks for the evidence-first survey pipeline: citation health, per-section coverage, placeholder leakage, and template repetition.

Write structured notes for each paper in the core set into `papers/paper_notes.jsonl` (summary/method/results/limitations).

Merge/simplify an over-fragmented outline to hit a paper-like section budget (NO PROSE): target final ToC ~6–8 H2, fewer thicker H3.

Rewrite subsection openers so they stop reading like a generated table-of-contents: remove \\\"overview/narration\\\" stems and reduce repeated opener cadences across H3s.

Scaffold a LaTeX project (`latex/main.tex`, optional bibliography wiring, structure) from an existing Markdown draft.

Compile a LaTeX project and run basic QA (missing refs, bib errors, broken citations), producing `latex/main.pdf` and a build report.

Write one survey-quality paragraph from evidence packs (tension → contrast → evaluation anchor → limitation).

Global consistency review for survey drafts: terminology, cross-section coherence, and scope/citation hygiene.

Write the survey's front matter files (Abstract, Introduction, Related Work, Discussion, Conclusion) in paper voice, with high citation density and a single evidence-policy paragraph.

Use when `evidence-review` has screened includes and needs a schema-aligned extraction table.

Evidence self-loop for surveys: read evidence bindings + evidence packs, then write an actionable upstream TODO plan (which stage/skill to fix) before writing more prose.

Create per-subsection evidence packs (NO PROSE): claim candidates, concrete comparisons, evaluation protocol, limitations, plus citation-backed evidence snippets with provenance.

Use when `paper-review` needs a claim-by-claim evidence gap report grounded in an extracted claim ledger.

Audit and rewrite evaluation/numeric claims to ensure they carry minimal protocol context (task + metric + constraint) and avoid underspecified model naming.

Audit-style editing pass for `output/DRAFT.md`: remove template boilerplate, improve coherence, and enforce citation anchoring.

Use when a review workspace has manuscript text and needs a traceable claim ledger.

Seed `outline/claim_evidence_matrix.md` from outline+notes; claim-matrix-rewriter rewrites it later from evidence packs.

PRODUCES a citation-budget plan/report (`output/CITATION_BUDGET_REPORT.md`); does NOT edit the draft. Allocates an in-scope per-H3 budget (NO NEW FACTS) so drafts stop failing the global unique-citation gate; citation-injector applies it.

Regression-check citation anchoring (citations stay in the same subsection) to prevent “polish drift” that breaks claim→evidence alignment.

Write H2 chapter lead blocks (`sections/S<sec_id>_lead.md`) that preview the chapter's comparison lens and connect its H3 subsections, without adding new facts.

Build per-chapter (H2) writing briefs (NO PROSE) so the final survey reads like a paper (chapter leads + cross-H3 coherence) without inflating the ToC.

Use when `evidence-review` has an extraction table and needs lightweight risk-of-bias fields.