Documentation Index
Fetch the complete documentation index at: https://docs.medera.info/llms.txt
Use this file to discover all available pages before exploring further.
Category: Evidence & Safety · Signals: Clinical question · Evidence library · Patient context · Currency of evidence
Overview
The Evidence Synthesis Agent runs at the point of decision, searching an approved evidence library and synthesizing the evidence with explicit citations and currency. Conclusions are bounded to the evidence base — never invented, never extrapolated.How this agent works
Configuration
This agent takes the following configuration from your practice settings:- Approved evidence library (PubMed, UpToDate, internal)
- Citation requirements
- Currency thresholds
- Surfacing policy
Execution
At runtime, the agent executes the following steps:- Receives a clinical question with patient context
- Searches the approved evidence library
- Synthesizes the evidence with explicit citations
- Reports currency and bound the answer to documented sources
- Logs the query and the response for audit
Composable experts
The Evidence Synthesis Agent bundles the following experts from the registry:| Expert | Role |
|---|---|
| PubMed Expert | Peer-reviewed literature search |
| UpToDate Expert | Clinical reference search |
| Evidence Synthesis Expert | Synthesis with bounded conclusions |
| Citation Expert | Citation rendering and currency |
Typical use cases
- Surface peer-reviewed evidence at the point of decision
- Cite every clinical claim with currency
- Bound conclusions to the evidence base
- Log queries and responses for audit
Compliance & safety guarantees
- Every claim carries an explicit citation
- Currency of evidence is reported
- Conclusions are bounded to documented sources
Tags
Evidence · PubMed · Synthesis · Cited
Related
Agentic Framework Overview
Learn how agents compose with experts and the orchestrator.
Expert Registry
Browse the typed expert registry agents bundle.
Quickstart
Send your first message to an agent.
A2A Protocol
Learn how agents talk to each other.