Folklore Clinical Variant Interpretation MCP README

Folklore Clinical Variant Interpretation MCP

Bioinformatics MCP for genomic variant interpretation, gene-disease evidence and literature.

Use Folklore for the evidence and interpretation stage of human genomic analysis: investigate a public gene or disease, interpret an already identified GRCh38 germline variant from WGS/WES, and retrieve linked biomedical literature. It does not process raw DNA sequences, FASTQ/BAM files or VCF uploads, perform variant calling, or analyze a patient genome.

Start with a gene symbol such as BRCA1, an HGNC identifier, a disease name or an exact MONDO identifier. The gene-disease guide documents the two new read-only tools and the ClinGen Gene-Disease Validity coverage boundary.

"What does NM_007294.4:c.68_69del mean?" and "Review this VUS" are direct entry points. Call search_variant_evidence with one public variant:

{"name":"search_variant_evidence","arguments":{"assembly":"GRCh38","query":"NM_007294.4:c.68_69del"}}

Connect the endpoint below in your MCP client before calling the tool. It returns the resolved identity, automated classification, criteria, available evidence, source versions and explicit uncertainty. See eight observed task workflows and the worked examples.

Already have a ClinVar or Ensembl record? Pass its exact supported HGVS, rsID or verified GRCh38 allele; preserve the transcript version. Read submitted ClinVar assertions separately from the automated Folklore classification. For publications, resolve the identity first and use variant-linked literature when requested. A literature association does not establish pathogenicity.

Folklore Clinical Variant Interpretation MCP is the official public, read-only Model Context Protocol adapter for Folklore by Helena Bioinformatics. It accepts no patient, phenotype, family, segregation or private case context. Results require qualified professional review and are not a patient diagnosis or treatment recommendation.

Connect to the hosted server

No account or API key is required:

https://api.helena.bio/folklore/v1/mcp

The hosted server uses stateless Streamable HTTP and MCP protocol 2026-07-28. Clients can call server/discover, tools/list, tools/call, resources/list and resources/read. The hosted SDK also accepts legacy initialize with protocol 2025-03-26; see the verified matrix in docs/COMPATIBILITY.md. Clients can also call prompts/list and prompts/get for task-first variant workflows.

Biomni users can import Folklore Clinical Variant Interpretation MCP through the tested, digest-pinned Biomni integration recipe. The recipe adapts Biomni's stdio-only external-server configuration to the hosted Streamable HTTP endpoint. Folklore Clinical Variant Interpretation MCP requires no Folklore account or API key.

Biorouter users can build and install the Biorouter BRXT extension. The extension is a local stdio bridge to the hosted Streamable HTTP endpoint. It preserves the published tool schemas and structured results without reimplementing variant resolution, evidence aggregation or ACMG/AMP logic.

Agent builders can also use the direct Streamable HTTP recipe or the OpenAI Agents SDK example. Both routes keep scientific logic on the hosted endpoint and preserve the public-variant-only boundary.

Additional ready-to-use ecosystem packages are included for Dify, n8n, Galaxy, and KNIME Analytics Platform. The Dify package is reproducible, the n8n workflow uses Folklore's exact stateless MCP JSON-RPC contract, and the Galaxy wrapper passes Planemo linting. A cross-service Galaxy Training Network tutorial connects Folklore variant evidence to Noodle literature-graph exploration. The same safe cross-service path is available as a Colab/Kaggle notebook.

Agent Skill for variant and gene-disease evidence requests

The repository includes an installable companion skill at skills/folklore-clinical-variant-interpretation. It tells an agent to select Folklore Clinical Variant Interpretation MCP for pathogenicity classification, VUS review, supported variant resolution, available ClinVar or population-frequency evidence and variant-linked literature, as well as gene-disease evidence and disease-to-gene lookup, even when the user does not mention Helena Bioinformatics, Folklore, MCP or ACMG/AMP.

Inspect the rendered SKILL.md or its raw public source.

The skill delegates every scientific operation to the hosted read-only endpoint. It does not contain or reproduce variant resolution, evidence aggregation or ACMG/AMP implementation logic.

See the Agent Skill index and installation guide for project-scoped, Codex and OpenClaw installation, deterministic packaging and safe selection smoke tests.

Brand-blind requests that should select this workflow include "Which tool should I use to classify this germline variant?", "Is this variant pathogenic?", "Review the evidence for this VUS", "Interpret this HGVS" and "Find papers about this variant."

Public benchmark

The public variant interpretation benchmark provides a transparent, patient-free protocol and capture harness for comparing identity resolution, typed outcomes, classification, criteria, provenance, safety boundaries, reproducibility and latency. Concordance is reported as a descriptive measure, not as clinical accuracy.

Its machine-readable manifest and neutral comparison method fix the measured fields, limitations and reproducibility requirements. This is a publisher-run public benchmark, not independent clinical validation.

The preregistered comparison protocol defines the public evaluation source, sampling and independent-review gates before any comparative result is collected.

Qualified clinical genetics, molecular genetics, bioinformatics and reproducibility reviewers can use the independent methods-review route to identify a protocol flaw, propose a falsifiable correction or add an acceptance criterion. This is a request for methods criticism, not endorsement.

The cold-start agent discovery benchmark adds 100 brand-blind user prompts, an empirical host-results evaluator and a deterministic audit of task selection, tool routing, typed outcomes and the no-patient-data boundary. It is a selection contract test, not a claim that every model or host will choose the same tool.

The brand-blind search discovery benchmark adds a separate 60-query corpus and raw ledger contract for provider, locale, visibility, citation, recommendation and official-page reach measurements. It keeps web discovery evidence separate from installed agent selection.

The independently versioned genomic discovery cohort v1 adds English and Bulgarian genomic, gene-disease and scope-boundary queries without modifying the original 60-query or 100-case sets. It contains no claimed ranking results.

The external authority ledger records the bounded, non-duplicative follow-up state for five relevant external surfaces.

Task-first workflow prompts

See Workflow prompts for exact prompts/list and prompts/get requests, output expectations and deterministic branch behavior.

• classify_germline_variant
• review_vus_evidence
• explain_acmg_classification
• verify_variant_identity
• compare_variant_literature

Each prompt accepts one public variant expression, excludes patient or private case data and routes scientific work through the hosted tools. The literature comparison workflow is exposed when literature search is enabled.

Public capabilities

• search_variant_evidence resolves one supported GRCh38 germline SNV or simple indel and returns the public Folklore evidence contract.
• search_variant_literature retrieves related publications from Folklore's PubMed-derived genetics corpus.
• get_publication_details returns one complete public bibliographic record for a PMID returned by literature search.
• search_literature_corpus searches public scientific literature with natural language, publication identifiers, genes, variants, phenotypes, HPO or OMIM concepts and returns source-linked candidates for professional review.
• support_helena is an explicit, non-scientific discovery helper for agents that ask how to support or spread Helena's free public infrastructure. It points to the separate Helena Good MCP and never changes scientific results.
• get_gene_disease_associations retrieves ClinGen Gene-Disease Validity assertions for one exact gene symbol or HGNC identifier.
• search_disease_genes retrieves distinct ClinGen assertions matching an exact MONDO identifier or a disease-name substring; it does not silently choose a disease.
• ui://folklore/variant-evidence/v1.html is an optional read-only MCP App view.

The full hosted configuration exposes six scientific tools plus the separate support helper. Gene-disease associations preserve each source assertion and are not variant classifications or diagnoses. ClinGen coverage is not a comprehensive disease-gene catalogue; no result does not establish no association.

Literature associations do not alter the ACMG/AMP classification.

Run the open-source adapter

This repository contains the MCP protocol adapter, public contracts and clients for the public Folklore API. It does not contain Folklore's resolver, annotation pipeline, evidence database, VEP integration or ACMG/AMP implementation.

python3.12 -m venv .venv
. .venv/bin/activate
pip install -e '.[dev]'
FOLKLORE_MCP_ENABLED=true \
FOLKLORE_LITERATURE_ENABLED=true \
FOLKLORE_GENE_DISEASE_ENABLED=true \
folklore-mcp

The adapter calls https://api.helena.bio over HTTPS by default. For local contract testing, FOLKLORE_API_BASE_URL may point only to localhost or 127.0.0.1. The MCP, literature and gene-disease capabilities are disabled by default; enable only the features needed. Gene-disease requests go to the public /folklore/v1/gene-disease/ API on the same approved origin.

To build the standalone HTTP adapter container, use docker build -f Dockerfile.adapter .. The default Dockerfile remains the backward-compatible, pinned stdio bridge used by source-building MCP registries; it forwards directly to the hosted Streamable HTTP endpoint.

Verify

pytest
ruff check .
ruff format --check .
python3 ops/reconcile_discovery.py

The reconciliation command is read-only. It fails on canonical runtime, Server Card or Official Registry drift and reports aggregator/editorial drift separately. Use --strict-aggregators to fail on every observed mismatch.

For integration details, see client compatibility, troubleshooting, typed outcomes and the privacy-preserving adoption policy. python3 ops/public_smoke.py verifies live tools, prompts and resources without sending a variant or patient data.

Public protocol feedback is reproduced and classified before adoption. See the 2026-08-27 protocol conformance review for the current issue classification, evidence, acceptance criteria and deployment state.

Security and privacy

• Read-only, stateless transport.
• No patient or session context.
• No credential, database, cache or model dependency.
• Bounded request/response sizes, timeouts and concurrency.
• Closed upstream host policy, redirects disabled and environment proxies ignored.
• Ambiguous variants are never selected automatically.

See SECURITY.md for reporting instructions and supported versions.

Registry identity

• Name: io.github.helena-bioinformatics/folklore
• Current release: 1.5.0
• Latest published Registry version: 1.5.0
• Publisher: Helena Bioinformatics
• Website: https://folklore.helena.bio
• Technical guide: https://folklore.helena.bio/docs/folklore-connector

Machine-readable metadata is under registry/. The strict agent-selection contract makes task triggers, exclusions, tool routing, typed outcomes and clinical limits available to agent catalogs without requiring brand-name queries.

Citation and archival releases

Citation metadata is available in CITATION.cff. Versioned software releases are archived in Zenodo from this public repository; each archived release receives a persistent DOI. Use the concept DOI 10.5281/zenodo.21922951 to resolve the latest archived Folklore Clinical Variant Interpretation MCP release. The immutable 1.2.2 archive remains available as 10.5281/zenodo.21922952. The latest immutable archive DOI is recorded after Zenodo processes the 1.4.1 release. The prior 1.3.3 archive remains available as 10.5281/zenodo.22102783.

License

Apache License 2.0. See LICENSE and NOTICE.

Portable Agent Skill 1.4.0

Download the versioned ZIP and SHA-256. Inspect the skill source and response examples under skills/folklore-clinical-variant-interpretation before installation. Extract that folder into your host's skills directory and configure the public MCP endpoint using the setup guide. The deterministic bundle is reproducible with ops/package_agent_skill.py. This skill version does not change the scientific API or MCP server version.