Pith MCP: arXiv paper reviews and claim checks in your AI app

Connect Pith's paper reviews to Claude, ChatGPT or Cursor

1. 1Click Connect Pith free. There's no sign-in, so its tools are ready at once.
2. 2Pith's server is read-only. Some tools are aliases of others, like `get_claims` for `analyze_paper_claims`; switch off the duplicates if you'd like your AI app to see a shorter list.
3. 3Add ContextGate to your AI app as a connector or MCP server. The steps for each app are on the Works with page.
4. 4Paste an arXiv link and ask your AI app for Pith's verdict, then open Activity to see the `resolve_paper` and `get_review` calls.

Research questions for Pith

• Get Pith's review of one paper with `get_review`: the verdict, confidence, novelty and correctness scores and its weakest assumption.
• See which papers support or contradict a statement with `check_claim`.
• Verify that a cited paper says what your draft attributes to it with `verify_citation`.
• Find a paper's references and the papers citing it with `citation_neighbors`.

Pith's server has 24 tools; all of them appear in ContextGate as soon as you add it.

A record of every paper your AI app checked

• Switch off any of the 24 tools; it's hidden from your AI apps and refused if called.
• Policies check every call, so a secret pasted along with a draft can be blocked before it reaches Pith.
• Activity records each review, claim check and citation lookup.

What you get with Pith and ContextGate

• Your AI apps can use PithClaude, Cursor, ChatGPT and other AI apps reach Pith through the one ContextGate connection you add to them.
• You choose which tools they can useSwitch off any Pith tool you don't want used. It's hidden from your AI apps and refused if called.
• Every call is checked and recordedYour policies check each call (for example, personal data is hidden and secrets are blocked), and Activity keeps a record of it.

ContextGate sees only the tool calls that go through it, not your AI app's built-in tools or your chats.

Pith tools

Examples of what your AI apps can do in Pith. You'll see the exact list of tools after you connect it.

• get_reviewPith's review of one specific paper, in one call: the verdict, confidence, novelty and correctness scores, one-line summary, weakest assumption, and the
• search_papersSearch Pith-reviewed papers by title and abstract. Call this when the user names a research topic, phenomenon, method, or paper title without an identif
• latest_reviewsThe papers Pith reviewed most recently, newest first. Call this when the user asks what is new, notable, or recently reviewed in a field or overall. day
• analyze_paper_claimsThe structured argument of one specific paper: what it actually asserts, the premises it assumes, the machinery connecting those premises to the claim,
• get_claimsAlias of analyze_paper_claims: the same ledger of what the paper asserts, its premises, machinery, and falsifier, in argument order.
• search_claimsFull-text search over every scientific claim Pith has extracted, across all reviewed papers. Call this when the user asks who claims something, whether
• get_paper_falsifierThe falsifier for one specific paper's core claim: the concrete observation, experiment, or computation that would settle or disprove it, plus how many
• get_falsifierAlias of get_paper_falsifier: the same falsifier and watcher count.
• get_signed_reviewsPublished human signed reviews of one paper: reviewer identity, axis deltas against the machine review, claim verdicts, defects, predictions, and who pa
• get_reviewer_recordA reviewer's published signed reviews and, when present, their calibration record (Brier score, open predictions, positivity skew). Call this when the u
• get_recordThe compiled record of one paper: the authors' own sentences, arranged with typed labels. It carries the main claim in the authors' words, each abstract
• resolve_paperTurn a pasted arXiv id, arXiv URL, DOI, or Pith Number into the arXiv id the other tools key on. ``resolved`` is false when the input did not become an
• paper_packetOne paper's compiled record, small enough to hold beside others: the main claim and contribution in the authors' words, each abstract sentence with its
• citation_neighborsOne hop from a paper: every reference entry with its resolved work identity, the papers that cite it, and, where a record exists, the citing sentences w
• paper_bridgeWhether either paper cites the other, the resolved works both cite (on or off arXiv), and the papers that cite both.
• paper_connectionsWhat other arXiv papers do with this paper, and what it does with the arXiv papers it cites: builds on, uses the method or tool, uses the data, compares
• check_claimWhich arXiv papers support a scientific statement, which contradict it, and which study it without deciding, each with the paper's own main claim quoted
• verify_citationWhether the cited paper's own sentences support what a sentence attributes to it: supports, partly supports (a condition is missing), contradicts, or no
• find_passagesAuthor sentences the classifier labeled with one role, across every stored record, each with its probability and paper. Abstract-tier records cover limi
• record_coverageSources read and levels run for one paper's record, or the number of stored records by tier. Call before treating an empty result as a finding: an empty