New studies, retractions, and NIH/FDA updates publish every week, and a frozen model sees none of them. Connect Claude, ChatGPT, or Cursor to a curated, cited evidence graph, updated in reviewed releases.
https://aviado.ai/api/mcpStart immediately with 10 anonymous tool calls per week—no account or API key required. Sign in anytime to unlock 25 calls/week via OAuth or a 30-day Bearer token.
https://aviado.ai/api/mcp as the Remote MCP Server URL.https://aviado.ai/api/mcp
https://aviado.ai/api/mcp. All 11 tools, 3 resources, and 3 prompts.Generate a free Authorization: Bearer <token> header at aviado.ai/oauth/authorize to bump headless CLI & IDE clients from 10 to 25 calls/week.
Science publishes every week. Frontier model weights are frozen until the next training run.
Ask for "blood pressure" and Aviado returns status: "needs_confirmation" with data: null and the candidates (systolic or diastolic) instead of silently picking one.
{
"status": "needs_confirmation",
"data": null,
"candidates": [
{
"id": "8462-4",
"display_name": "Diastolic Blood Pressure"
},
{
"id": "8480-6",
"display_name": "Systolic Blood Pressure"
}
]
}found: false never means "safe."Every pair check reports dataset coverage (none | sparse | moderate | broad) for the FDA drug-label record it searched, so a pair with no record is flagged for pharmacist review, never reported as safe.
{
"verdict": {
"status": "not_found",
"found": false
},
"coverage": {
"level": "sparse",
"label_rows": 1
}
}Biomarker results carry prior_weight (0–1 graph evidence weight, not effect size), study and RCT counts, study-mean dose context and PubMed PMIDs. Condition results carry grades (A–D). Products carry ProofMark (0–100), a quality score, not efficacy.
{
"display_name": "Berberine",
"prior_weight": 0.996,
"graph_status": "proven",
"n_studies": 27,
"rct_count": 12,
"dose": {
"value": 1070,
"unit": "mg"
},
"pmids": [
"41543854",
"40439602",
"39998703"
]
}Supplements in the evidence graph that move a biomarker (e.g. LDL cholesterol), with direction, prior_weight, study counts, study-mean dose and PMIDs.
e.g. “What supplements have the strongest human evidence for lowering LDL cholesterol?”
Mixed interventions (supplements, foods, drugs and more) that move a biomarker, each labelled with its intervention_type.
Biomarkers a supplement moves in the graph, strongest edge per biomarker, ordered by prior_weight.
e.g. “Does ashwagandha lower cortisol?”
Overview of one supplement: uses, top 5 biomarkers with up to 3 PMIDs each, mechanism tags and cited safety contraindications.
Interaction profile from FDA drug labels plus curated supplement–supplement pairs. Sparse for many supplements.
Point check of one pair (supplement or medication): verdict, severity, evidence level and coverage. Not found is not safe.
e.g. “Can I take St. John's wort with sertraline?”
Best product per brand for a supplement, ranked by ProofMark (0–100 product quality, not efficacy).
e.g. “Which vitamin D3 products have the highest ProofMark scores?”
Resolve a supplement name to its Aviado id. Exact matching, no fuzzy guessing.
Resolve a health condition, symptom or goal to its Aviado condition id.
Supplements studied for a condition or goal, strongest evidence first, with grades A–D. Proxy rows are labelled as hypotheses.
e.g. “Which supplements are studied for insomnia?”
Starter stack (default 3, max 5) for a goal, screened for supplement–supplement interactions. Drug interactions are checked only for medications you list in exclude.
e.g. “Build me a starter stack for LDL cholesterol.”
How Aviado's evidence is graded and shown: visibility, evidence weight and grades, proxy rows, citations, doses as context, identity and interaction sources.
What status values, candidate relations, evidence grades, severity scales, coverage levels and interaction relations mean.
Which interactions and evidence Aviado holds and where it is sparse or absent: interaction sources, matching rules, evidence gaps. Absence is not safety.
Orchestrate search, overview, biomarker evidence and interactions for one supplement, with safety framing and coverage caveats.
Resolve each supplement and medication, check every pair with check_interaction, and summarise with coverage caveats.
Find the supplements in Aviado's evidence that raise or lower a biomarker, with citations and caveats.