Harvest
Retrieve the complete Official MCP Registry and search GitHub repositories, code, curated catalogues, and citation chains.
Methods appendixProtocol v0.2-draft
How we search
the landscape.
A living, reproducible protocol for finding MCP servers, connectors, skills, libraries, and agents relevant to real-world evidence.
Last updated July 10, 2026 · Metadata verification snapshot completed.
Search architectureRecall first.
Recall first.
Judgment second.
Discovery pairs a technology signal with a health-workflow concept. The complete Official MCP Registry is also harvested and screened locally, so projects do not need to use our terminology to be found.
Discover
Run a reproducible matrix of technology signals against controlled RWE, HEOR, public-health, method, data, disease, and exposure terms.
Screen
Classify each candidate as direct RWE, enabling infrastructure, adjacent, excluded, duplicate, unavailable, or flagged.
Verify
Record the canonical source, implementation evidence, maintenance signal, provenance, limitations, and exact discovery path.
Positive control 01PopHIVE
If a search wave cannot recover a known relevant project, screening stops and the discovery logic is revised.
Controlled vocabularyThe terms we use
The terms we use
to find the work.
Terms are searched as acronyms, expanded forms, plurals, and predictable variants. Broad words such as “disease,” “connector,” “FDA,” or “bias” must be paired with a stronger technology or RWE signal.
01
Technology & packaging
17 terms- MCP
- Model Context Protocol
- MCP server
- MCP tool
- MCP connector
- connector
- FastMCP
- @modelcontextprotocol/sdk
- mcp.server
- server.json
- mcp.json
- SKILL.md
- Claude skill
- agent skill
- plugin
- agent
- tool server
02
RWE, HEOR & effectiveness
26 terms- real-world evidence
- RWE
- real-world data
- RWD
- health economics and outcomes research
- HEOR
- health economics
- outcomes research
- comparative effectiveness
- treatment effectiveness
- vaccine effectiveness
- cost-effectiveness
- economic evaluation
- budget impact
- QALY
- DALY
- patient-reported outcome
- pharmacoepidemiology
- epidemiology
- population health
- public health
- burden of disease
- natural history
- incidence
- prevalence
- surveillance
03
Safety & pharmacovigilance
21 terms- pharmacovigilance
- drug safety
- adverse event
- adverse drug reaction
- signal detection
- FAERS
- FDA Adverse Event Reporting System
- VAERS
- Vaccine Adverse Event Reporting System
- EudraVigilance
- VigiBase
- WHO-UMC
- ICSR
- E2B
- MedDRA
- disproportionality
- PRR
- ROR
- EBGM
- post-marketing
- PASS
04
Regulatory & product data
21 terms- FDA
- openFDA
- EMA
- MHRA
- PMDA
- Health Canada
- TGA
- ICH
- CIOMS
- regulatory intelligence
- regulatory science
- labeling
- package insert
- drug approval
- biologic
- medical device
- recall
- 510(k)
- orphan designation
- EPAR
- REMS
05
Healthcare & public-health data
27 terms- electronic health record
- EHR
- EMR
- claims
- administrative data
- patient registry
- pharmacy
- dispensing
- laboratory
- mortality
- death certificate
- vital statistics
- hospitalization
- wearable
- social determinants of health
- CDC WONDER
- NHANES
- BRFSS
- NVSS
- NCHS
- SEER
- WHO
- GBD
- outbreak
- biosurveillance
- wastewater surveillance
- vaccine coverage
06
Models, standards & terminology
20 terms- OMOP
- OHDSI
- common data model
- CDM
- FHIR
- HL7
- CDISC
- SDTM
- ADaM
- SNOMED
- ICD-9
- ICD-10
- RxNorm
- LOINC
- CPT
- NDC
- UMLS
- terminology mapping
- phenotype
- cohort definition
07
Tokenization, linkage & engineering
18 terms- tokenization
- privacy-preserving tokenization
- patient token
- record linkage
- data linkage
- probabilistic linkage
- deterministic linkage
- privacy-preserving record linkage
- PPRL
- patient matching
- entity resolution
- deduplication
- interoperability
- ETL
- data harmonization
- data integration
- identity resolution
- crosswalk
08
Study design, causal inference & bias
33 terms- study design
- observational study
- causal inference
- target trial
- target trial emulation
- estimand
- g-methods
- g-formula
- g-computation
- g-estimation
- marginal structural model
- inverse probability weighting
- time-varying confounding
- confounding
- unmeasured confounding
- confounding by indication
- propensity score
- difference-in-differences
- instrumental variable
- negative control
- quantitative bias analysis
- sensitivity analysis
- selection bias
- information bias
- misclassification
- immortal time bias
- collider bias
- DAG
- adjustment set
- self-controlled case series
- case-crossover
- new-user design
- active comparator
09
Evidence generation & synthesis
16 terms- clinical trial
- ClinicalTrials.gov
- literature search
- PubMed
- systematic review
- scoping review
- meta-analysis
- evidence synthesis
- PRISMA
- citation verification
- evidence grading
- protocol
- statistical analysis plan
- Table 1
- survival analysis
- time-to-event
10
Diseases, populations & exposures
33 terms- disease
- chronic disease
- infectious disease
- communicable disease
- autoimmune
- immune-mediated
- cardiovascular
- oncology
- cancer
- diabetes
- metabolic
- respiratory
- neurologic
- psychiatric
- renal
- hepatic
- rare disease
- maternal
- pediatric
- geriatric
- pregnancy
- vaccine
- vaccination
- immunization
- treatment
- therapy
- medication
- drug
- device
- biologic
- diagnostic
- exposure
- intervention
Decision rules
Indexed does not mean validated.
Eligible
A public MCP server, connector, agent skill, library with an agent-facing interface, plugin, or orchestrated agent workflow that supports an identifiable RWE, HEOR, epidemiology, public-health, evidence-synthesis, or enabling activity.
Not eligible
Generic health assistants, incidental MCP mentions, nonfunctional demonstrations, duplicates, implausible bulk projects, or projects connected only by a broad disease or treatment word.
Stopping rule
Every predefined query is run, the Official MCP Registry is fully screened, two synonym and citation-chaining rounds produce no new eligible projects, and every candidate has a recorded disposition.