Touch Browser vs Scholar Feed MCP

Page-local claim verdicts

Confidence bands and review recommendations

Citation URLs with retrieval timestamps

Reusable evidence snippets

Multi-page session synthesis

Headless browser research

search_papers

Search Scholar Feed's 600k+ CS/AI/ML paper corpus. Defaults to semantic (embedding) search — finds conceptually related papers even when the user's wording doesn't match the paper's title/abstract. Pass mode='keyword' for exact-string full-text search. CAVEAT: semantic search often misses old high-citation CANONICAL papers (e.g. foundational anchors like H2O for KV eviction, GRIT for unified embedding+generation) because the ranker prefers recent stylistically-matched papers. If you're hunting the canonical anchor for an area, parse the top-5 result abstracts for baseline mentions ('we compare against X, Y, Z'), then look the most-mentioned name up directly. Returns papers with LLM-generated summaries, novelty scores, and structured extraction data. Default response is a lean 13-field shape (arxiv_id, title, authors, year, categories, has_code, github_url, citation_count, venue_name, llm_summary, llm_significance, llm_novelty_score, impact_pct) — pass verbose=true or fields=... for the full shape with method/task/dataset extraction. RANKING BY IMPACT — two different notions, don't confuse them: (1) PROVEN impact = citations. For 'the important/seminal papers on topic X', pass sort='impactful' (most-cited among the relevant) or sort='balanced' (relevant AND well-cited). This is the right tool for established/foundational work. (2) FORECAST impact = impact_pct (0-100), an ML percentile of PREDICTED citations computed WITHIN a paper's own arXiv-category cohort over the last ~90 days — so it is a cohort-relative rank, not an absolute score: impact_pct=100 means 'top of its category this window', which in a quiet category is a much weaker claim than in a busy one. (An A+/A/B/C/D impact_tier field was previously returned alongside it; it was removed because a single global cutoff ladder could not be calibrated honestly across both fresh and mature papers. Do not expect it, and do not treat any letter grade you may have cached as current.) For 'what's rising/new in X' pass sort='trending' or filter impact_min=N — but NOTE impact_pct is NULL on everything older than ~90 days, so impact_min DROPS all established/canonical papers (it is NOT a way to find the influential papers in a niche — use sort='impactful' for that). Both impact notions are distinct from llm_novelty_score (new-idea-ness, an orthogonal filter). (3) ADOPTION impact = GitHub traction. Pass sort='community' to rank by real-world engineering adoption (stars + star-velocity) — the papers practitioners are actually running/building on, a signal independent of citations. But NOTE its coverage: github_stars is 0-defaulted, so an unmeasured repo is INDISTINGUISHABLE from a genuinely unstarred one (it reads as 0 stars, not NULL), and star coverage is heavily skewed toward recently-published papers — most older papers with a repo have never had stars fetched, so their real popularity is invisible here. sort='community' therefore buries established/canonical work rather than ranking it low on merit. Treat a 0 as 'unknown', not 'unpopular', and use sort='impactful' or min_citations for older work. Coverage is being backfilled, so this skew shrinks over time; never infer a paper is unadopted from a 0. Filter on it with min_stars=N (minimum GitHub stars) and has_code=true (only papers with a code release); has_code/min_stars surface RUNNABLE/ADOPTED work, the engineering counterpart to citations. github_url_exists=true is the stricter has_code (requires a linked repo). Supports filtering by category, novelty, recency, method, task, dataset, and contribution type — plus min_citations (minimum PROVEN citations, keeps established papers unlike the ~90-day impact_min) and an explicit date window via published_after / published_before ('YYYY-MM-DD', vs days' rolling lookback). v3 ABSORPTIONS: pass sort='trending' to rank by rising/forecast impact (impact_pct); pass anchor_paper_id to replicate find_similar (q is ignored in anchor mode, results carry similarity_score); pass scope_to_citations_of to restrict search to a paper's citation graph (replaces find_citations_about).