AnswerThis MCP

Connect AnswerThis to your assistant

Updated November 11, 2026 · 59 tools

AnswerThis lets your AI assistant search academic papers, work with the papers in your AnswerThis library, create and edit cited Writer drafts, and run systematic reviews, from screening and extraction to risk of bias, GRADE and the write-up. You don't need an account to read this guide or to search public papers; other tools ask you to sign in to AnswerThis.

Connect

Pick your assistant below. The AnswerThis MCP page has the same options as one-click cards.

Claude

Open AnswerThis in the Claude directory and add it. On Claude Team and Enterprise plans, owners (or Enterprise roles allowed to manage connectors) add it; members then select Connect in Customize → Connectors.

Can't use the Claude directory? Add AnswerThis as a custom connector and choose Sign in when needed under Authentication on step 2. Claude may preselect No sign-in, and then it won't ask you to sign in when a tool needs your AnswerThis account.

ChatGPT

Open AnswerThis in the ChatGPT plugin directory and select Install plugin.

Codex

Codex uses the same plugin: install it from the plugin directory, or in the Codex CLI run codex plugin add answerthis@openai-curated-remote.

Cursor

Add AnswerThis to Cursor. The link opens Cursor and adds the server to its MCP config.

Claude Code and other MCP clients

Add the server by URL: https://api.answerthis.io/mcp. For a JSON config such as Cursor's mcp.json or Claude Code's .mcp.json:

{
  "mcpServers": {
    "answerthis": {
      "type": "http",
      "url": "https://api.answerthis.io/mcp"
    }
  }
}

Then sign in to AnswerThis from the client (in Claude Code, run /mcp).

Sign in and approve

Public paper search (search_papers) and product help (get_answerthis_docs) work without an account. Every other tool, and searching your own library, needs your AnswerThis account.

The first time a tool needs your account, your assistant opens AnswerThis sign-in. Sign in or create an account, check which assistant is asking and the permissions it requests, then approve. Your assistant receives a connection token, not your password. Never paste a password, API key or token into a chat.

Library, full text, Writer drafts, systematic reviews and jobs are separate permissions. Reading full text depends on whether the source is available to you; AnswerThis does not bypass publisher access.

Available tools

These are the 59 tools the AnswerThis server offers. Your assistant's tool list is authoritative for their arguments. “Not used” means the tool does not use your plan's API calls or processing credits.

Tool What it does Needs an account? API allowance
Papers and help
search_papers Search by topic, DOI, title or author, with citations and filters. Public search: no. Library: yes. Not used
get_evidence_passages Find citable abstract passages for a question or selected papers. Yes Not used
get_answerthis_docs Read help for setup, research, your library and Writer. No Not used
Library
list_library_folders Find the folders in your reading library. Yes Not used
list_library_papers Browse the papers saved in a folder. Yes Not used
save_papers Save selected papers to your library. Yes Not used
read_library_paper Read accessible full text from your library, one page at a time. Yes Not used
Writer
search_drafts Find your saved Writer documents. Yes Not used
read_draft Read a saved draft and its citations. Yes Not used
create_draft Save a new Writer draft with linked citations. Yes Not used
update_draft Edit an existing draft and preserve its citation links. Yes Not used
upload_document Queue a PDF or DOCX with a downloadable file URL for import into Writer. Yes Not used
get_import_status Check whether an import is processing, completed or failed. Yes Not used
Systematic reviews
create_slr_session Create a systematic review in a project you can access. Yes Not used
get_slr_session Read a review's saved inputs and revision. Yes Not used
edit_slr_session Change a review's saved inputs; reviewer decisions and history are kept. Yes Not used
count_slr_search Count results for the review's saved searches in each database. Yes 1 API call
confirm_slr_search Confirm counted searches before retrieving papers. Yes Not used
start_slr_fetch Retrieve the papers found by a confirmed search. Yes 1 API call
resume_slr_step Screen or extract papers that still need it, reusing valid results, or run risk of bias, meta-analysis, GRADE or the write-up. Yes Screening and extraction: 1 API call + credits per paper. Other stages: included in SLR-enabled plans
update_slr_review Find new papers and continue a saved review without removing existing work. Included in SLR-enabled plans. Yes Not used
get_slr_update Read an update's progress, database coverage, errors and resulting version. Yes Not used
list_slr_updates List the saved updates for a review. Yes Not used
confirm_slr_update Confirm an update paused after finding more than 5,000 new papers. Yes Not used
cancel_slr_update Stop further update work; added papers and saved results are kept. Yes Not used
list_slr_versions List immutable versions of a saved review. Yes Not used
get_slr_version Read a saved review version, including its draft and evidence summary. Yes Not used
get_slr_version_changes Compare review versions: new papers, decisions, synthesis changes and work needing review. Yes Not used
get_slr_risk_of_bias Read each included paper's risk-of-bias ratings and justifications. Yes Not used
list_slr_meta_analyses List saved meta-analyses, with any held for the owner's judgement. Yes Not used
get_slr_grade Read GRADE certainty ratings and whether they are still current. Yes Not used
get_slr_write_up Summarize the current manuscript draft; read its full text over REST. Yes Not used
get_slr_prisma Read the review's current PRISMA flow counts. Yes Not used
list_slr_executions List a review's search, retrieval and processing runs. Yes Not used
get_execution_status Check the progress of one run. Yes Not used
list_execution_jobs List the screening or extraction jobs a run started. Yes Not used
list_execution_sources Show count and retrieval progress for each query and database. Yes Not used
cancel_execution Stop a run from starting more work; finished results are kept. Yes Not used
list_slr_papers List the papers in a review and whether they are selected. Yes Not used
get_slr_results Read current screening and extraction results with their evidence, or past attempts. Yes Not used
list_slr_screening_decisions Read current screening decisions, effective judgments and versions for editing. Yes Not used
update_slr_screening_decisions Set, reconfirm or revert screening decisions while preserving review history. Yes Not used
create_slr_snapshot Save a stable screening snapshot for comparing research updates. Yes Not used
get_slr_snapshot Read a saved screening snapshot and its counts. Yes Not used
list_slr_snapshot_results Read the papers, decisions and evidence saved in a screening snapshot. Yes Not used
list_slr_snapshot_changes Compare two saved screening snapshots and read their changes. Yes Not used
Screening and extraction jobs
start_screening Start a job that screens a set of papers. Yes 1 API call + credits per paper
start_extraction Start a job that extracts study, arm or outcome data from papers. Yes 1 API call + credits per paper
get_job_status Check a job's progress and credit use. Yes Not used
get_job_results Read a job's results and errors, item by item. Yes Not used
list_jobs List your screening and extraction jobs. Yes Not used
cancel_job Stop a job from scheduling more work; finished items are kept. Yes Not used
retry_job Retry a finished job's failed items as a new linked job. Yes 1 API call + credits per paper
Papers and uploads for jobs
add_papers Add up to 100 papers by metadata, DOI, PMID or PDF link. Yes Not used
get_paper Read a paper's metadata and whether its text is ready. Yes Not used
get_paper_content Read a paper's text, or the exact source a job item used. Yes Not used
create_uploads Create upload slots for PDFs or CSV/JSONL job inputs. Yes Not used
complete_uploads Ask AnswerThis to verify finished uploads. Yes Not used
get_upload_status Check whether an upload has been verified. Yes Not used

API calls and credits

Most tools are not counted against your plan's API allowance. The tools that start counts, fetches, screening or extraction use API calls, and screening and extraction also use processing credits per paper, as the table shows.

start_screening, start_extraction, retry_job and resume_slr_step accept max_credits to cap what that work can spend. Sign in to the AnswerThis MCP page and open Usage to see your use.

Disconnect and support

Tool results, including private papers and drafts you let your assistant read, are sent to that assistant. See the MCP privacy supplement and connector terms.

To disconnect, sign in to the AnswerThis MCP page and select Disconnect on the assistant's card or under Other connected apps. That connection can no longer use your AnswerThis account, and your saved work stays. A client you connected more than once, such as Cursor on two computers, can be listed once per connection; disconnect each one. For help, email [email protected] or use the contact page.