AdisInsight
Query Springer Nature's drug-pipeline and trial database from an AI agent.
Search the NIH/NLM clinical trial registry from your AI agent.
Opens Agent Studio, where connecting is one click. The connector URL below works in any MCP client.
Paste it into any MCP client. Setup docs
The Clinical Trials connector puts the NIH/NLM clinical study registry inside Claude, ChatGPT and any MCP-compatible agent. Its six read-only tools search ClinicalTrials.gov by condition, drug, location, sponsor, investigator and recorded eligibility criteria, and pull full protocol records by NCT ID. Anthropic builds and hosts it. It needs no sign-in.
Verified connector
Listed by Anthropic as an Anthropic-built connector in its Connectors Directory.
Connection checked by Agentman on .
Anthropic states this reflects the level of review a connector received, not a security audit.
Search ClinicalTrials.gov database for clinical trials. This is the PRIMARY tool for finding trials. WHEN TO USE: - Finding trials for a disease/condition (e.g., 'What trials exist for lung cancer?') - Finding trials testing a specific drug/treatment (e.g., 'Find pembrolizumab trials') - Finding trials in a geographic area (e.g., 'Clinical trials in Boston') - General trial discovery and research questions USE DIFFERENT TOOLS FOR: - Detailed info on a specific trial by NCT ID -> use get_trial_details - All trials by a specific company -> use search_by_sponsor (or search_trials with sponsor=...) - Analyzing endpoints/outcomes -> use analyze_endpoints - Patient eligibility matching -> use search_by_eligibility QUERY SYNTAX (for condition, intervention, sponsor, location): - Boolean: 'cancer AND immunotherapy', 'aspirin OR ibuprofen', 'tumor NOT benign' - Exact phrase: '"breast cancer"' (with quotes) - Grouping: '(lung OR breast) AND cancer' - Synonyms are automatically included (e.g., 'heart attack' finds 'myocardial infarction') BEST PRACTICES: - Start with condition parameter for disease-focused searches - Add status=['RECRUITING'] to find active trials patients can join - Use phase filter for specific development stages (PHASE1, PHASE2, PHASE3, PHASE4) - Set count_total=true to know total matches (useful for: 'How many trials exist for X?') - Use page_size=50-100 for broader overviews, page_size=10 for quick lookups
Get comprehensive details for a specific clinical trial using its NCT ID. WHEN TO USE: - User provides a specific NCT ID (e.g., 'Tell me about NCT04567890') - Need full eligibility criteria, endpoints, or locations for a specific trial - Following up on a trial found via search_trials - Answering detailed questions about a known trial - Verifying patient eligibility for a specific trial USE search_trials INSTEAD FOR: - Finding trials (this tool requires knowing the NCT ID) - Browsing trials by condition/intervention/sponsor WHAT THIS RETURNS: - Full eligibility criteria (inclusion/exclusion) - Study design and methodology - Primary, secondary, and other endpoints with timeframes - All study locations with contact info - Sponsor and collaborator details - Study dates and enrollment numbers - Results link if trial has published results NCT ID FORMAT: 'NCT' followed by 8 digits (e.g., NCT04567890, NCT00001234)
Find all clinical trials sponsored by a specific company or organization. Functionally equivalent to search_trials(sponsor=...). WHEN TO USE: - Questions about a company's pipeline (e.g., 'What is Pfizer working on?') - Competitive intelligence (e.g., 'What cancer drugs is Novartis developing?') - Tracking pharma company portfolios and development programs - Finding trials from academic institutions (e.g., 'Mayo Clinic', 'NIH') USE search_trials INSTEAD FOR: - Disease-focused searches where sponsor doesn't matter - Finding trials by treatment name rather than sponsor EXAMPLES: - 'Pfizer Phase 3 trials' -> sponsor_name='Pfizer', phase=['PHASE3'] - 'Moderna COVID vaccines' -> sponsor_name='Moderna', condition='COVID-19' - 'Active Merck oncology trials' -> sponsor_name='Merck', condition='cancer', status=['RECRUITING'] TIPS: - Partial names work: 'Pfizer' matches 'Pfizer Inc', 'Pfizer Pharmaceuticals' - Set count_total=true to get total number of trials by sponsor - Combine with phase filter to see early vs late stage pipeline
Find principal investigators (PIs) and research sites conducting trials in a therapeutic area. WHEN TO USE: - 'Who are the leading researchers in Alzheimer trials?' - 'Find investigators at Mayo Clinic working on cancer' - 'Which sites in California are running diabetes trials?' - Site selection for planning new trials - Building investigator networks and collaborations USE search_trials FOR: - Finding trials themselves rather than investigators - When you need trial details, not investigator info WHAT THIS RETURNS: - Investigator names and roles (Principal Investigator, Sub-Investigator) - Institutional affiliations - Facility/site names - Geographic locations - Associated trial NCT IDs and titles TIPS: - Use condition parameter to focus on a disease area - Add institution to find investigators at specific hospitals/universities - Use location for geographic focus (city, state, country) - Increase page_size to 50-100 for more comprehensive investigator lists - Use investigator_name for direct name search via advanced query syntax ADVANCED: For direct investigator name search, use the investigator_name parameter which searches in OverallOfficialName and ResponsiblePartyInvestigatorFullName fields.
Analyze primary and secondary outcome measures (endpoints) from clinical trials. MODES (provide ONLY nct_id OR condition, not both): 1. Single Trial: Provide nct_id ONLY to analyze one specific trial's endpoints 2. Aggregate: Provide condition ONLY to analyze patterns across multiple trials If both provided, nct_id takes precedence (single trial mode). SINGLE TRIAL MODE (nct_id): - Returns all endpoints for the specified trial - Useful for understanding specific trial design - Example: nct_id='NCT03661411' AGGREGATE MODE (condition): - Analyzes endpoints across many trials in a therapeutic area - Identifies common endpoint patterns and measures - Useful for protocol design and competitive analysis - Example: condition='diabetes', phase=['PHASE3'] WHAT THIS RETURNS: - List of primary endpoints (main efficacy measures) - List of secondary endpoints (additional outcomes) - List of other endpoints (exploratory outcomes) - Most common measures across analyzed trials - Timeframes for each endpoint measurement EXAMPLES: - Single trial: nct_id='NCT03661411' - Phase 3 cancer endpoints: condition='cancer', phase=['PHASE3'] - Recent diabetes outcomes: condition='diabetes', start_date_after='2022-01-01'
Find clinical trials matching specific patient eligibility criteria. Use this for patient-trial matching and finding trials a specific patient might qualify for. DEFAULT STATUS: Only searches RECRUITING trials. To include completed, upcoming, or all trials, explicitly set the status parameter. WHEN TO USE: - Patient matching: 'Find trials for a 65-year-old female with diabetes' - Specific criteria: 'Trials requiring HbA1c > 8%' or 'BRCA positive trials' - Age-restricted searches: 'Pediatric cancer trials' or 'Trials for elderly patients' - Finding trials by inclusion/exclusion criteria USE search_trials FOR: - General disease/condition searches - When patient demographics don't matter ELIGIBILITY KEYWORDS TIPS: - Use medical abbreviations: 'ECOG', 'HbA1c', 'BMI', 'eGFR' - Search criteria text: 'prior chemotherapy', 'treatment naive' - Biomarkers: 'BRCA mutation', 'HER2 positive', 'PD-L1' - Lab values: 'creatinine', 'ALT', 'bilirubin' - ICD-10 codes indicate specific subtypes (E10.x=Type 1, E11.x=Type 2, etc.) - Disease subtypes matter: Type 1 vs Type 2 diabetes, HER2+ vs HER2- cancer, etc. EXAMPLES: - '65yo diabetic patient' -> condition='diabetes', min_age='18 Years', max_age='70 Years' - 'Breast cancer with BRCA' -> condition='breast cancer', eligibility_keywords='BRCA' - 'Recruiting trials for men with prostate cancer' -> condition='prostate cancer', sex='MALE'
Read from the server on 2026-08-17, including each tool's own safety annotations.
get_trial_details reports only whether results were posted.search_by_eligibility takes only a condition, age bounds, a sex value and criteria keywords. Do not paste identifiable health information into a prompt in order to run a search — the connector has nowhere to put it and no need for it.initialize and tools/list handshake. Everything above about tool behaviour comes from the server's own descriptions and schemas plus Anthropic's published tutorial. The registry counts come from ClinicalTrials.gov's own API, not from this connector.anthropics/life-sciences plugin manifest points clinical-trials at a deepsense.ai host rather than the directory endpoint, and we could not handshake that host. See the setup section above.No, and no tool on this server is built to. search_by_eligibility filters registry listings by the criteria sponsors recorded — age bounds, sex, and keywords in the criteria text. It never sees you, your records or your labs, so it cannot assess a person. ClinicalTrials.gov itself says to discuss options with your health care professional before joining any study.
Anthropic lists it at anthropic tier in its Connectors Directory and hosts the endpoint on its own infrastructure at hcls.mcp.claude.com. The data is not Anthropic's. Every result comes from ClinicalTrials.gov, the registry the National Library of Medicine maintains, where study sponsors and investigators submit and own the records.
No. We opened an anonymous handshake against the endpoint on 2026-08-17 and it returned all six tool definitions with no credential. The directory records the server as authless, the host publishes no OAuth protected-resource descriptor, and ClinicalTrials.gov states its data are available to all requesters at no charge.
No. Registration and recruitment are separate facts. Of 598,890 studies in the registry on 2026-08-17, ClinicalTrials.gov counted 65,002 as recruiting — roughly one in nine. Pass status RECRUITING to search_trials to filter, and note that search_by_eligibility already defaults to recruiting trials only unless you override it.
Because results reporting is required only for some studies and lags by design. ClinicalTrials.gov reported 79,715 studies with posted results on 2026-08-17, about 13 percent of the registry. The Final Rule at 42 CFR Part 11 covers FDA-regulated drug, biological and device trials other than Phase 1, so much of the registry falls outside it.
No. It reads the outcome measures sponsors registered — the measure name, its time frame and its description — then groups and counts them across the trials it analysed. It reports which endpoints are common in a therapeutic area. It does not read result data, compare efficacy, or say whether any endpoint was met.
No. Anthropic's directory carries two ClinicalTrials.gov servers. This page describes the anthropic-tier Clinical Trials connector at hcls.mcp.claude.com with the six tools below. A separate community-tier listing, ClinicalTrials.gov Explorer by BLEN, runs on different infrastructure with a disjoint three-tool surface. Community connectors are excluded from this catalogue.
claude.com/robots.txt is Allow: / with no Content-Signal line (retrieved 2026-08-17). · retrieved 2026-08-17initialize and tools/list handshake against https://hcls.mcp.claude.com/clinical_trials/mcp — six tools with descriptions, input and output schemas and annotations, over an anonymous connection (2026-08-17). serverInfo reported clinical_trials version 3.2.0. Source for the tool count, every tool description and parameter named on this page, the uniform read-only/non-destructive/ idempotent annotations, the single openWorldHint: false on get_trial_details, and the server's own instructions naming ClinicalTrials.gov API v2 as its backing service. No tool was called. · retrieved 2026-08-17https://hcls.mcp.claude.com/.well-known/oauth-protected-resource returned 404 (retrieved 2026-08-17), consistent with the successful anonymous handshake and with the directory's authless record. robots.txt and llms.txt on that host also 404. · retrieved 2026-08-17/api/v2/version returned API version 2.0.5 with a data timestamp of 2026-08-17T09:00:05; /api/v2/stats/size returned 598,890 total studies; study queries filtered on overallStatus=RECRUITING and on results availability returned 65,002 and 79,715 respectively (all retrieved 2026-08-17). These are the registry's own live counts, read directly rather than through the connector. clinicaltrials.gov/robots.txt disallows /api/ generally while explicitly allowing /api/int/ and /api/seo/; the counts above come from the documented public /api/v2/ data API, whose use the site's terms permit at no charge, fetched as single directed reads rather than crawled. · retrieved 2026-08-17README.md and .mcp.json return 404. · retrieved 2026-08-17www.anthropic.com/robots.txt is Allow: / with no Content-Signal line (retrieved 2026-08-17). · retrieved 2026-08-17anthropic tier, the life-sciences category, is_authless: true, Read permissions, and Anthropic as author. The same snapshot carries a separate community-tier ClinicalTrials.gov Explorer entry by BLEN with the three tools search_api, describe_schema and execute. · retrieved 2026-01-07Connect Clinical Trials once and your agents call these tools on their own: on a schedule, in a workflow, with nobody at the keyboard.