Skip to content

HypePaper MCP server

app.hypepaper/hypepaperhypepaper.appstreamable-httpv2.0.0Scanned

State of the art, trending papers, benchmarks, citations and topic experts from HypePaper.

58 tools: 45 change things and 13 only read.

Tools: 0 delete, 0 irreversible, 45 write, 13 read.
  • Reads private data · 2
  • Reads text others write · 3
  • Can send data out · 2
  • No findings

Tools listed without sign-in. Remote posture

With Anlyon

45 tools here change things. Govern them.

  • Anlyon lists the tools and adopts none.
  • Each tool you adopt gets approvals and receipts.
  • Your agent never holds the key.
Govern this server →

A tool your agent calls directly is not routed through Anlyon and is not governed by it. Anlyon Radar only lists and labels, and these labels are the same either way.

What the HypePaper MCP server can do

The HypePaper MCP server is published by hypepaper.app, a domain the MCP Registry verified. It has 58 tools. It can create or change data (45 tools: trending_papers, new_sota, account, expert_notes and 41 more), but cannot delete data. 13 tools only read.

2 tools read private data (list_views and fetch); 3 tools bring in text other people can write (api_get_llm, list_views and fetch); 2 tools can send data out of your system (list_views and fetch). On its own it has all three parts of a prompt-injection exfiltration path: private data, text others write, and a way out.

None of the fixed checks matched anything in its tool definitions. That is not a safety guarantee.

Its remote endpoint lists its tools without sign-in.

Last scanned Nov 14, 2026, version 2.0.0.

Tools

Grouped by what each tool does, most consequential first. Open one for its reasons and full description; descriptions are the publisher's own words.

58 tools

Changes things 45Claude Code: ask
  • accountYour plan, credits remaining this month, and upgrade URL.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Your plan, credits remaining this month, and upgrade URL. Call this when a request was denied. [Free · 1 credit]
  • affiliation_profileA lab or company's output: paper counts, GitHub stars and 30-day delta, top authors with h-index, recent papers.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    A lab or company's output: paper counts, GitHub stars and 30-day delta, top authors with h-index, recent papers. [Free · 2 credits]
  • api_getCall one read-only HypePaper API route from api_catalog as yourself and return its JSON.No findings

    Why it is in this group

    • The name "api_get" says "get": a word for reading.
    • The description's first verb is "call": a word for a change.
    Description, as the server wrote it
    Call one read-only HypePaper API route from api_catalog as yourself and return its JSON. Free responses are cut at 24 KB. [Free · 1 credit]
  • api_get_llmCall one LLM-backed HypePaper API route (explain, tldr, translate, lit-review, briefing, …) from api_catalog and return its JSON.Reads text others writeNo findings

    Why it is in this group

    • The name "api_get_llm" says "get": a word for reading.
    • The description's first verb is "call": a word for a change.

    What it can reach

    • untrusted-inputThe description's first sentence says "call" and "review": it reads comments, which other people can write.
    Description, as the server wrote it
    Call one LLM-backed HypePaper API route (explain, tldr, translate, lit-review, briefing, …) from api_catalog and return its JSON. [Free · 5 credits + AI credits]
  • authorAn author's profile by name or id: affiliations (curated table first, else the newest paper's first page), metrics (papers, citations, GitHub stars, active years, h-index when know…No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    An author's profile by name or id: affiliations (curated table first, else the newest paper's first page), metrics (papers, citations, GitHub stars, active years, h-index when known), their 10 most-cited papers, top 10 collaborators with shared-paper counts, and `more` hints for author_papers / coauthor_network. Use when a question names a person; use search to find the exact name first. A name several people share returns no figures, only who those people are (`ambiguous`, with each one's `openalex_id`): ask again with one. A name nobody is told apart under yet answers by name, marked `name_only`. [Free · 1 credit]
  • author_papersEvery paper under an author's name, paginated and sorted by date, citations, GitHub stars or hype, each with venue, citations, stars and co-authors.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Every paper under an author's name, paginated and sorted by date, citations, GitHub stars or hype, each with venue, citations, stars and co-authors. Use after `author` when the top-10 list is not enough or you need the chronological output. A name several people share returns no figures, only who those people are (`ambiguous`, with each one's `openalex_id`): ask again with one. A name nobody is told apart under yet answers by name, marked `name_only`. [Free · 1 credit]
  • benchmark_timelineSOTA progression on one dataset x metric (optionally one protocol/condition) over time: the record-setting papers in order with value, delta over the previous record and how long i…No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    SOTA progression on one dataset x metric (optionally one protocol/condition) over time: the record-setting papers in order with value, delta over the previous record and how long it held, plus every dated entity as a point. Use for 'how has X improved on Y', 'when did the record last change' or to plot a benchmark. Select a condition key from dataset_leaderboard when multiple protocols exist. A record whose jump is outside the SOTA feed's believability band is flagged believable=false (likely mis-extracted). [Free · 3 credits]
  • citation_graphVerified in/out citations for a paper with match scores and the matched reference text.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Verified in/out citations for a paper with match scores and the matched reference text. [Free · 1 credit]
  • citation_walkMulti-hop citation traversal (up to 3 hops) from a paper, forward (papers citing it), backward (its references) or both, as nodes with in-corpus citation counts and hop depth, and …No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Multi-hop citation traversal (up to 3 hops) from a paper, forward (papers citing it), backward (its references) or both, as nodes with in-corpus citation counts and hop depth, and edges carrying the fuzzy match score and the matched reference text. Use to trace lineage, find follow-up work, or build a reading graph; use citation_graph for a flat one-hop page. Inbound edges come newest first, outbound competitors first, so a truncated walk is the top of each side. [Free · 2 credits]
  • coauthor_networkCo-authorship graph around an author: hop 1 is everyone they published with (top 20 by shared papers), hop 2 adds the collaborators of their 8 closest co-authors.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Co-authorship graph around an author: hop 1 is everyone they published with (top 20 by shared papers), hop 2 adds the collaborators of their 8 closest co-authors. Nodes carry hop depth, edges the shared-paper count. Use to map a research group, find who bridges two labs, or pick reviewers. Consortium papers with more than 30 authors are skipped on hop 2. For a person, every node is a person with its `openalex_id`. A name several people share returns no figures, only who those people are (`ambiguous`, with each one's `openalex_id`): ask again with one. A name nobody is told apart under yet answers by name, marked `name_only`. [Free · 2 credits]
  • cocitation_recommendationsPapers most often cited alongside the given paper(s) by the papers that cite them — the neighbours keyword search misses because they use different terminology.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Papers most often cited alongside the given paper(s) by the papers that cite them — the neighbours keyword search misses because they use different terminology. Returns each candidate with its co-citation frequency, citations, hype and code flag. Use for 'what else should I read with X' or to grow a reading list from 1-20 seed papers. [Free · 1 credit]
  • conference_papersAccepted papers of one venue-year (2020 onwards: ICML, NeurIPS, ICLR, CVPR, ...), most cited first, paginated, with acceptance status (oral/poster/spotlight), primary area, citatio…No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Accepted papers of one venue-year (2020 onwards: ICML, NeurIPS, ICLR, CVPR, ...), most cited first, paginated, with acceptance status (oral/poster/spotlight), primary area, citations, hype, code and project URLs; optional keyword filter on title/abstract and an acceptance-type filter. Use for 'what was at CVPR 2026 on X', to build a venue reading list, or with status='oral' for the papers the programme singled out. [Free · 1 credit]
  • dataset_leaderboardLeaderboards for a dataset: every metric board with top methods and their papers, plus recent papers, the paper that introduced the dataset, and papers per year with the first and …No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Leaderboards for a dataset: every metric board with top methods and their papers, plus recent papers, the paper that introduced the dataset, and papers per year with the first and last year. Each board has a condition key and label; rankings compare matching protocols only. [Free · 4 credits]
  • expert_askAsk any topic's expert a question (a topic without its own expert is answered from its nearest ancestor's, else from its own papers).No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Ask any topic's expert a question (a topic without its own expert is answered from its nearest ancestor's, else from its own papers). The answer is written from the expert's knowledge graph, paper views, SOTA lanes, benchmark rows and notes; each sentence ends with [Cn] markers that `citations` resolves (hp:// pointer, kind, title, the quoted passage, the paper's URL), and it says what the evidence does not establish instead of guessing. [Free · 5 credits + AI credits]
  • expert_communitiesLouvain communities of a topic expert's knowledge graph at one dendrogram level, largest first, with each community's hub nodes and a sample of members resolved to names/types.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Louvain communities of a topic expert's knowledge graph at one dendrogram level, largest first, with each community's hub nodes and a sample of members resolved to names/types. Use to get the shape of a field before drilling in: level 0 is the finest split, max_level the coarsest. Then expert_node_context on a hub. [Free · 1 credit]
  • expert_federationWhich topic experts hold the same entity — a paper (by arXiv id), method, model, dataset, benchmark or metric — and under which node id in each.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Which topic experts hold the same entity — a paper (by arXiv id), method, model, dataset, benchmark or metric — and under which node id in each. Pass a name or a node id; topic_id narrows the starting expert. Use to move a question to the expert that knows an entity best, or to see how widely a method is used across fields. [Free · 1 credit]
  • expert_graph_mapA budgeted subgraph around a query: seeds are the best full-text hits in the expert's graph, then neighbours join by how many edges tie them to what is already held, until `budget`…No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    A budgeted subgraph around a query: seeds are the best full-text hits in the expert's graph, then neighbours join by how many edges tie them to what is already held, until `budget` nodes. Returns nodes, the edges among them and the papers they came from. Use for 'what surrounds X in this field' in one call; then expert_node_context to go deeper. [Free · 2 credits]
  • expert_node_contextOne node of an expert's graph with all its edges (direction, type, evidence text), its neighbours and their papers; depth 2 adds the neighbours' edges.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    One node of an expert's graph with all its edges (direction, type, evidence text), its neighbours and their papers; depth 2 adds the neighbours' edges. Use after expert_search_facts / expert_graph_map / expert_communities gave you a node id. [Free · 1 credit]
  • expert_notesCurated notes of a topic expert (kind, title, content, evidence paper ids) without an LLM call.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Curated notes of a topic expert (kind, title, content, evidence paper ids) without an LLM call. Free shows the first 300 chars per note. [Free · 1 credit]
  • expert_verify_claimDeterministic check of a (subject, relation, object) claim against an expert's graph — no LLM.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Deterministic check of a (subject, relation, object) claim against an expert's graph — no LLM. Verdicts: supported (a direct edge, matching relation when given), supported_via (a direct edge of another relation, or a 2-hop chain), contradicted (a contradicting edge), unknown (no path within 2 hops; NOT a refutation). Returns the evidence chain with edge evidence text and the source papers. Subject/object accept a node id, an exact name, or search words. [Free · 2 credits]
  • expert_wikiThe expert's wiki: one structural page per public node (papers, concepts, entities, topics, questions, syntheses) with its description and every relation in and out.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    The expert's wiki: one structural page per public node (papers, concepts, entities, topics, questions, syntheses) with its description and every relation in and out. Without slug: list pages (filter with q) and counts per kind. With slug: the page body. Use for a readable dossier on one thing; use expert_search_facts to find claims. [Free · 1 credit]
  • lane_leadersState of the art of a topic BY GRAPH TRAVERSAL: for every benchmark lane (dataset × metric × experimental protocol) in the topic's expert graph, the paper at the top of the `better…No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    State of the art of a topic BY GRAPH TRAVERSAL: for every benchmark lane (dataset × metric × experimental protocol) in the topic's expert graph, the paper at the top of the `better` chain — the one that beats others and is beaten by none — with the edge evidence (values compared). Use for 'what is SOTA on <topic>' when you want the answer as graph structure; sota_papers gives fuller rows. [Free · 3 credits]
  • library_claims_for_paperYour personal library: Return all claim-type kg_nodes (PerformanceClaim, LimitationClaim, ComparisonClaim, ...) outbound from a Paper kg_node id.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Your personal library: Return all claim-type kg_nodes (PerformanceClaim, LimitationClaim, ComparisonClaim, ...) outbound from a Paper kg_node id. [Free · 1 credit]
  • library_emergent_topicsYour personal library: List emergent Trend / ApproachFamily clusters ordered by paper_count (default) or created_at when 'since' is provided.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Your personal library: List emergent Trend / ApproachFamily clusters ordered by paper_count (default) or created_at when 'since' is provided. Use for 'what's trending?' or 'what's new in this area?' style questions. Returns cluster name, type, paper_count, and a preview of member concepts. [Free · 1 credit]
  • library_filterYour personal library: Restrict a list of node_ids to those connected to a paper in the requesting user's tracked-paper library.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Your personal library: Restrict a list of node_ids to those connected to a paper in the requesting user's tracked-paper library. Use when the user asks about 'my papers' or 'my library'. [Free · 1 credit]
  • library_node_contextYour personal library: Fetch a node's neighborhood subgraph (nodes + edges) at the given depth.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Your personal library: Fetch a node's neighborhood subgraph (nodes + edges) at the given depth. Use to expand context around a specific node id you've already located. [Free · 1 credit]
  • library_papers_with_conceptYour personal library: Papers connected to the Concept node id (any direction).No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Your personal library: Papers connected to the Concept node id (any direction). [Free · 1 credit]
  • library_papers_with_methodYour personal library: Papers that use the given Method node id.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Your personal library: Papers that use the given Method node id. [Free · 1 credit]
  • library_statsYour personal library: Summary statistics about the requesting user's tracked-paper library: total paper count, top years, top venues, recent titles, most-starred, AND most_connect…No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Your personal library: Summary statistics about the requesting user's tracked-paper library: total paper count, top years, top venues, recent titles, most-starred, AND most_connected_papers (top-5 by kg_edge degree — answers 'which paper has the most connections'). Use FIRST whenever the user asks anything about THEIR library size, distribution, recent additions, or graph connectivity. Always library-scoped — no node_id inputs needed. [Free · 1 credit]
  • library_synthesize_papersYour personal library: Generate a multi-paper synthesis.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Your personal library: Generate a multi-paper synthesis. Pass paper UUIDs (the papers.id row id, NOT the kg_nodes id) and an optional synthesis prompt. Returns a prose summary and saves it to the user's library as a re-readable artifact. [Free · 1 credit + AI credits]
  • library_timelineYour personal library: Chronological list of papers matching a free-text query.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Your personal library: Chronological list of papers matching a free-text query. Use for 'how has X evolved' style questions. [Free · 1 credit]
  • new_sotaPapers from the last 21 days that beat the incumbent on a benchmark, with dataset, metric, value and the incumbent.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Papers from the last 21 days that beat the incumbent on a benchmark, with dataset, metric, value and the incumbent. [Free · 3 credits]
  • paper_fulltextFull text of a paper (extracted text, else OCR markdown, else abstract), paginated by character offset.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Full text of a paper (extracted text, else OCR markdown, else abstract), paginated by character offset. [Free · 5 credits]
  • paper_referencesA paper's resolved reference list, paginated, best-matched first: each entry is the corpus paper it resolved to (id, arXiv id, date, venue, citations), the reference's own title/au…No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    A paper's resolved reference list, paginated, best-matched first: each entry is the corpus paper it resolved to (id, arXiv id, date, venue, citations), the reference's own title/authors/year as printed, the raw reference text, and how it was resolved (match method, score, verified). Use to audit what a paper builds on or to pick seeds for citation_walk. [Free · 1 credit]
  • paper_resultsEvery result row extracted from one paper's tables — its own numbers and the baselines it reports — each with its SOTA standing on that lane: rank, number of entities, whether it i…No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Every result row extracted from one paper's tables — its own numbers and the baselines it reports — each with its SOTA standing on that lane: rank, number of entities, whether it is the current leader, the leader and the gap to it. Use to verify a paper's claims or to see which benchmarks it competes on before calling benchmark_timeline. Unknown experimental conditions return comparison_status=insufficient_information and sota=null. When rows_truncated is true, total_rows says how many exist: pass dataset to get one benchmark's rows, or a larger limit — do not fill the gap from memory. Rows do not carry the paper's table numbers; do not cite a Table number you have not read in the paper itself. [Free · 3 credits]
  • similar_papersNearest papers by embedding for a paper id.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Nearest papers by embedding for a paper id. [Free · 1 credit]
  • sota_alarmStanding alarm on state of the art: watch a topic or a benchmark lane, then poll for the record changes you have not seen yet.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Standing alarm on state of the art: watch a topic or a benchmark lane, then poll for the record changes you have not seen yet. action='watch' adds a watch (topic, or dataset with an optional metric), action='unwatch' removes one, action='poll' (the default) returns new events across every watch — the previous holder, the new leader, the delta, and how long the old record stood. A poll marks what it returned as seen and moves the cursor, so the next call is a true diff; pass peek=true to look without consuming. Use it to keep a project current instead of re-reading a leaderboard: one call a day tells you if anything moved. [Free · 10 credits]
  • sota_changelogBenchmark lanes whose leader changed since a date, corpus-wide or within a topic's subtree: for each change the previous holder, the new leader (paper, method, venue), the delta an…No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Benchmark lanes whose leader changed since a date, corpus-wide or within a topic's subtree: for each change the previous holder, the new leader (paper, method, venue), the delta and how long the old record stood. Use for 'what changed in SOTA on X since June' or a periodic catch-up. Examines the 150 most recently active lanes in scope; a change whose jump is outside the SOTA feed's believability band is not reported. [Free · 3 credits]
  • sota_checkIs this paper (or a dataset+metric+value) state of the art?No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Is this paper (or a dataset+metric+value) state of the art? Returns the current leader and who has beaten it since under the same experimental conditions. Missing protocol evidence or peers yields an unknown verdict. [Free · 8 credits]
  • sota_papersState of the art for a research topic, a dataset, or a free-text dataset query.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    State of the art for a research topic, a dataset, or a free-text dataset query. Use first for any 'what is SOTA on X', 'best method for X' or 'who leads benchmark X' question. Per benchmark lane (dataset x metric x condition) returns the leader, runners-up and entrants since `since` (default 365 days) with whether each beat the incumbent. Every row carries the value, its provenance (self-reported or reported as a baseline in another paper's table, match method), the paper (id, arXiv id, date, venue, citations, hype, code URL), up to 5 authors and, on paid plans, affiliations. Single-source lanes are flagged: one paper's table is not a leaderboard. Rankings require matching experimental protocols; rows carry conditions and source evidence. Pass the user's own words as `topic` when no topic name is known: scope.interpreted_as then names the topic and benchmarks it was read as (say so), and lists the newest matching papers with ranked_in_lanes — a paper without a lane row has no verified standing. [Free · 6 credits]
  • sota_walkWalk the `better` edges from a paper: direction 'up' follows papers that beat it (lane by lane) until the lane leader — i.e.No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    Walk the `better` edges from a paper: direction 'up' follows papers that beat it (lane by lane) until the lane leader — i.e. 'what is SOTA relative to this paper'; 'down' lists what it beats. Each step carries the compared values. Use after get_paper/search when you hold a paper id. [Free · 3 credits]
  • sota_with_codeState of the art you can actually RUN: per benchmark lane, the best-ranked paper that ships a code repository, with the repo (owner/name, stars, forks, watchers, description), its …No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    State of the art you can actually RUN: per benchmark lane, the best-ranked paper that ships a code repository, with the repo (owner/name, stars, forks, watchers, description), its rank among all entrants, and how much accuracy it gives up against the absolute leader. Use for 'best method with code for X', 'what can I reproduce today', or picking a baseline to build on. Two thirds of papers with benchmark numbers ship no code, so this answers a different question from sota_papers. Each repo carries link_confidence: 'verified' means the stored repo metadata cites the paper, 'unverified' means we could not tie them together — a linked repo is not proof the code runs. [Free · 10 credits]
  • topic_overviewOne taxonomy topic by name or id: where it sits (parents, level), its child topics with subtree paper counts, the benchmarks/datasets its papers report on, the topic experts availa…No findings

    Why it is in this group

    • Nothing in its name or description says what it does, so it is counted as a write.
    Description, as the server wrote it
    One taxonomy topic by name or id: where it sits (parents, level), its child topics with subtree paper counts, the benchmarks/datasets its papers report on, the topic experts available for expert_ask / expert_notes, and its top papers by hype. Use to orient in a field before sota_papers or to pick the right subtopic id. [Free · 1 credit]
Only reads 13Show 13 read-only tools
  • api_catalogList the read-only HypePaper API routes reachable through api_get / api_get_llm: path, summary, params, section and price.No findings

    Why it is in this group

    • The description's first verb is "list": a word for reading.
    Description, as the server wrote it
    List the read-only HypePaper API routes reachable through api_get / api_get_llm: path, summary, params, section and price. Filter by section or a substring. [Free · 1 credit]
  • compare_on_benchmarkCompare 2–5 papers head-to-head on the datasets and metrics they share; returns wins, losses and the evidence rows.No findings

    Why it is in this group

    • The name "compare_on_benchmark" says "compare": a word for reading.
    Description, as the server wrote it
    Compare 2–5 papers head-to-head on the datasets and metrics they share; returns wins, losses and the evidence rows. Only complete, matching experimental protocols are compared. coverage has one row per paper: comparison_status compared, no_shared_lane (its stored lanes are listed) or no_stored_result. A number not returned here is not stored; say so, never supply it. [Free · 8 credits]
  • expert_search_factsFull-text search over an expert's graph nodes — claims, results, methods, datasets, metrics, papers — ranked, with the source paper of each hit.No findings

    Why it is in this group

    • The name "expert_search_facts" says "search": a word for reading.
    Description, as the server wrote it
    Full-text search over an expert's graph nodes — claims, results, methods, datasets, metrics, papers — ranked, with the source paper of each hit. Filter with types (e.g. Paper, Method, Algorithm, Model, PerformanceClaim, Result). The entry point for node ids; cheaper than expert_ask when you want facts, not prose. [Free · 1 credit]
  • fetchRead one view by its hp:// pointer — from list_views, or from an expert_ask citation — sliced in the database to at most max_tokens (≤2,500): a TL;DR, key ideas (one by #index), a …Reads private dataReads text others writeCan send data outNo findings

    Why it is in this group

    • The name "fetch" says "fetch": a word for reading.

    What it can reach

    • private-dataThe description's first sentence says "read" and "database": it reads databases.
    • untrusted-inputThe input takes a URL (input.uri), and the tool reads it: it reads web pages, which anyone can write.
    • outboundThe input takes a URL (input.uri): a request to a URL can carry data to whoever runs it.

    Claude Code: ask, because it is tagged outbound.

    Description, as the server wrote it
    Read one view by its hp:// pointer — from list_views, or from an expert_ask citation — sliced in the database to at most max_tokens (≤2,500): a TL;DR, key ideas (one by #index), a deep-analysis section (analysis/<lang>/Q3, a subsection by #slug), result rows (results?ds=&metric=), a SOTA lane, a comparison, a note (current, or @v<history_id>), a story or synthesis section. Full text pages by offset. A missing view comes back empty with a reason. [Free · 1 credit]
  • get_paperPaper detail by UUID or arXiv id: abstract, authors, venue, hype and citation counts, GitHub stars, quick summary, benchmark rows and any SOTA takes.No findings

    Why it is in this group

    • The name "get_paper" says "get": a word for reading.
    Description, as the server wrote it
    Paper detail by UUID or arXiv id: abstract, authors, venue, hype and citation counts, GitHub stars, quick summary, benchmark rows and any SOTA takes. [Free · 1 credit]
  • library_compare_methodsYour personal library: Side-by-side comparison of methods: for each method id, list the papers that use it and outbound claim nodes.No findings

    Why it is in this group

    • The name "library_compare_methods" says "compare": a word for reading.
    Description, as the server wrote it
    Your personal library: Side-by-side comparison of methods: for each method id, list the papers that use it and outbound claim nodes. [Free · 1 credit]
  • library_find_contradictionsYour personal library: Find claims that contradict the input.No findings

    Why it is in this group

    • The name "library_find_contradictions" says "find": a word for reading.
    Description, as the server wrote it
    Your personal library: Find claims that contradict the input. Pass either a claim kg_node id (returns contradictions of that single claim) or a HypePaper papers.id (returns contradictions of any of the paper's claims). Walks 'contradicts_claim' edges in either direction. [Free · 1 credit]
  • library_find_extending_papersYour personal library: Find papers that extend or improve on the given paper's methods/algorithms.No findings

    Why it is in this group

    • The name "library_find_extending_papers" says "find": a word for reading.
    Description, as the server wrote it
    Your personal library: Find papers that extend or improve on the given paper's methods/algorithms. Use when the user asks 'what's been done since paper X?' or 'who extended this method?'. paper_id is the HypePaper papers.id (not the kg_nodes id). [Free · 1 credit]
  • library_get_paper_contentYour personal library: Fetch a paper's title, abstract, GitHub stats, AND its full deep-analysis markdown (the per-paper structured writeup with method, findings, numbers, limitati…No findings

    Why it is in this group

    • The name "library_get_paper_content" says "get": a word for reading.
    Description, as the server wrote it
    Your personal library: Fetch a paper's title, abstract, GitHub stats, AND its full deep-analysis markdown (the per-paper structured writeup with method, findings, numbers, limitations). Call this after `search_nodes` returns a Paper hit when you need to quote specific numbers, methods, or limitations from the paper rather than just the abstract. Returns the user's preferred-language version of the analysis (falls back to English). [Free · 1 credit]
  • library_search_nodesYour personal library: Search the research knowledge graph for nodes matching a free-text query.No findings

    Why it is in this group

    • The name "library_search_nodes" says "search": a word for reading.
    Description, as the server wrote it
    Your personal library: Search the research knowledge graph for nodes matching a free-text query. Returns enriched node payloads ({id, type, name, aliases, description, source_paper}) plus 1-hop edges around the top hits. Use this first when you need to find papers, methods, or concepts by topic. [Free · 1 credit]
  • list_expertsTopics that have a ready expert (name, layer, papers ingested, last built).No findings

    Why it is in this group

    • The name "list_experts" says "list": a word for reading.
    Description, as the server wrote it
    Topics that have a ready expert (name, layer, papers ingested, last built). Use before expert_notes / expert_ask. [Free · 1 credit]
  • list_viewsWhat HypePaper holds about a paper, topic, note, editor story or area synthesis, as hp:// pointers with token estimates, without the text: TL;DR, key ideas, methods, each deep-anal…Reads private dataReads text others writeCan send data outNo findings

    Why it is in this group

    • The name "list_views" says "list": a word for reading.

    What it can reach

    • private-dataThe description's first sentence says "note": it reads notes.
    • untrusted-inputThe input takes a URL (input.uri), and the tool reads it: it reads web pages, which anyone can write.
    • outboundThe input takes a URL (input.uri): a request to a URL can carry data to whoever runs it.

    Claude Code: ask, because it is tagged outbound.

    Description, as the server wrote it
    What HypePaper holds about a paper, topic, note, editor story or area synthesis, as hp:// pointers with token estimates, without the text: TL;DR, key ideas, methods, each deep-analysis section (Q1 problem … Q6 summary), result rows, SOTA brief and lanes, tables, figure captions, comparisons, notes. Pass hp://paper/<id>, hp://topic/<id>, hp://note/<id>, hp://story/<id>, hp://synthesis/<id>, or a paper UUID / arXiv id. Then fetch the views you need. [Free · 1 credit]

This is not a safety rating, and no finding is not a safety guarantee. The groups and findings come from fixed checks of the tool definitions the server returned on the date shown, not from its code or what it does when called. How the checks work.

Exfil surface

Which tools read private data, which read text someone else can write, and which can send data out. One of each is a path a prompt injection can use.

12 paths in this server alone from private data, through text someone else wrote, to a way out. Denying 2 outbound tools breaks every one.

Reads private data2
Reads files, mail, messages, databases, secrets, calendars or repositories.
Reads text others write3
Brings in text someone else can write: web pages, inbound email, messages, issues and comments.
Can send data out2
Can carry data to people or systems: it sends, posts or publishes, or takes a recipient, channel or URL.
deny to break every path
{
  "permissions": {
    "deny": [
      "mcp__hypepaper__fetch",
      "mcp__hypepaper__list_views"
    ]
  }
}

Other servers in the same agent add their own tools to these paths. Audit your whole setup to see them together.

Remote posture

What anyone on the internet sees at the remote endpoint: an anonymous initialize and tools/list, and the public OAuth metadata.

Tools listed without sign-in

Looked at Nov 15, 2026, 13:18 UTC, unchanged since Nov 15, 2026. No credentials were sent and no tool was called. What is looked at.

Reachable
yes
HTTP status of an anonymous initialize
200
tools/list served without credentials
yes
Tools that change things, listed without credentials
45
Protocol version
2025-06-18
WWW-Authenticate on 401
not observed
resource_metadata in WWW-Authenticate
not observed
RFC 9728 protected resource metadata
https://api.hypepaper.app/.well-known/oauth-protected-resource/api/v1/mcp
Authorization servers
https://zvesxmkgkldorxlbyhce.supabase.co/auth/v1
Legacy HTTP+SSE transport
no
Dynamic client registration
yes
Client ID metadata documents
no
  • WarnThe whole server

    The endpoint served tools/list without credentials, and the list has 45 tools that change things (45 write). Whether a call also works without credentials was not tested: tools are never called.

    Matched text

    https://api.hypepaper.app/api/v1/mcp: tools/list returned 58 tools to an anonymous client, including trending_papers (write), new_sota (write), account (write), expert_notes (write), similar_papers (write) and 40 more

    What this check looks for

Findings

Matches of fixed checks, with the text they matched.

No check matched anything in these tool definitions. That is not a safety guarantee.

History

Changes to its tool definitions and to its remote endpoint, newest first.

No change seen yet. Its tools are rescanned every day, and a remote endpoint is looked at once a week.

Install

Pinned to version 2.0.0. Values in angle brackets are yours to fill in.

claude mcp add --transport http hypepaper https://api.hypepaper.app/api/v1/mcp --header 'Authorization: <your-authorization>'
claude mcp add --transport http hypepaper https://api.hypepaper.app/api/v1/mcp --header 'Authorization: <your-authorization>'

Run in a terminal.

Claude Code permissions

Allow the reads, ask before a change or a way out, deny the deletes. Paste into .claude/settings.json.

.claude/settings.json
{
  "permissions": {
    "allow": [
      "mcp__hypepaper__search",
      "mcp__hypepaper__get_paper",
      "mcp__hypepaper__list_experts",
      "mcp__hypepaper__compare_on_benchmark",
      "mcp__hypepaper__api_catalog",
      "mcp__hypepaper__expert_search_facts",
      "mcp__hypepaper__library_search_nodes",
      "mcp__hypepaper__library_get_paper_content",
      "mcp__hypepaper__library_compare_methods",
      "mcp__hypepaper__library_find_extending_papers",
      "mcp__hypepaper__library_find_contradictions"
    ],
    "ask": [
      "mcp__hypepaper__trending_papers",
      "mcp__hypepaper__new_sota",
      "mcp__hypepaper__account",
      "mcp__hypepaper__expert_notes",
      "mcp__hypepaper__similar_papers",
      "mcp__hypepaper__citation_graph",
      "mcp__hypepaper__entity_trends",
      "mcp__hypepaper__dataset_leaderboard",
      "mcp__hypepaper__affiliation_profile",
      "mcp__hypepaper__expert_ask",
      "mcp__hypepaper__sota_check",
      "mcp__hypepaper__paper_fulltext",
      "mcp__hypepaper__sota_papers",
      "mcp__hypepaper__paper_results",
      "mcp__hypepaper__benchmark_timeline",
      "mcp__hypepaper__sota_changelog",
      "mcp__hypepaper__sota_with_code",
      "mcp__hypepaper__lane_leaders",
      "mcp__hypepaper__sota_walk",
      "mcp__hypepaper__citation_walk",
      "mcp__hypepaper__paper_references",
      "mcp__hypepaper__cocitation_recommendations",
      "mcp__hypepaper__author",
      "mcp__hypepaper__author_papers",
      "mcp__hypepaper__coauthor_network",
      "mcp__hypepaper__topic_overview",
      "mcp__hypepaper__conference_papers",
      "mcp__hypepaper__api_get",
      "mcp__hypepaper__api_get_llm",
      "mcp__hypepaper__expert_communities",
      "mcp__hypepaper__expert_federation",
      "mcp__hypepaper__expert_wiki",
      "mcp__hypepaper__expert_verify_claim",
      "mcp__hypepaper__expert_graph_map",
      "mcp__hypepaper__expert_node_context",
      "mcp__hypepaper__sota_alarm",
      "mcp__hypepaper__list_views",
      "mcp__hypepaper__fetch",
      "mcp__hypepaper__library_node_context",
      "mcp__hypepaper__library_stats",
      "mcp__hypepaper__library_filter",
      "mcp__hypepaper__library_claims_for_paper",
      "mcp__hypepaper__library_papers_with_method",
      "mcp__hypepaper__library_papers_with_concept",
      "mcp__hypepaper__library_timeline",
      "mcp__hypepaper__library_synthesize_papers",
      "mcp__hypepaper__library_emergent_topics"
    ],
    "deny": []
  }
}

Tools that only read are allowed, tools that change things ask first, tools that delete or cancel are denied. A read that can send data out (it takes a URL, say) asks first too. It assumes the server is installed as hypepaper. Review it before you use it.

Capability tags behind 3 rules
  • mcp__hypepaper__api_get_llm untrusted-input
  • mcp__hypepaper__list_views private-data, untrusted-input, outbound
  • mcp__hypepaper__fetch private-data, untrusted-input, outbound

Pin

Keep the decision: fail CI when the tools change, follow every change, or show a badge.

Pin in CI

Fail a build when the tools you reviewed are not the tools the server serves today.

curl -s "https://radar.anlyon.com/api/v1/servers/app.hypepaper/hypepaper/verify?hash=bbc759638cfbe3826667fd3240e9ce72363cf8a49e33002972eb527841c9c1e9" | jq -e '.changed == false'

Pinned hash bbc759638cfb, from Nov 14, 2026, 19:18 UTC.

Watch

Get told when its tools change: a new tool, a new description, a new class.

https://radar.anlyon.com/servers/app.hypepaper/hypepaper/changes.xml

An Atom feed. Works in any feed reader or Slack's RSS app.

Badge

Anlyon Radar badge for app.hypepaper/hypepaper
[![Anlyon Radar](https://radar.anlyon.com/badge/app.hypepaper/hypepaper.svg)](https://radar.anlyon.com/servers/app.hypepaper/hypepaper)

Source: MCP Registry. Synced Nov 14, 2026, 19:17 UTC.

HypePaper MCP server: questions and answers

Answered from the scan above, not from the publisher's description.

Is the HypePaper MCP server safe?

No scan can promise that. What its definitions show: 58 tools, 45 of which can change, send or delete something, and 0 findings from fixed checks. Allow the reads, ask before each change, and deny the deletes; the permissions block on this page does that.

Can the HypePaper MCP server delete data?

No tool is classed as destructive: going by their definitions, none of them deletes, cancels or revokes anything.

Can the HypePaper MCP server do things that cannot be undone?

No tool is classed as irreversible: none of them sends, pays, publishes or merges, going by their definitions.

What tools does the HypePaper MCP server have?

58 tools: 45 write, 13 read. trending_papers, new_sota, account, expert_notes, similar_papers, citation_graph, entity_trends, dataset_leaderboard and 50 more.

Does the HypePaper MCP server need sign-in or OAuth?

Its remote endpoint listed its tools to an anonymous client, without sign-in.

Does the HypePaper MCP server send data to outside services?

2 tools can carry data out (it sends, posts or publishes, or takes a recipient, channel or URL): list_views and fetch.

Has the HypePaper MCP server changed its tools?

Not since Anlyon Radar started tracking it. Last scanned Nov 14, 2026.