io.github.cyanheads/uniprot-mcp-server
v0.2.6Protein research over UniProtKB: search by function, fetch curated records, map IDs, proteomes.
Endpoint health
online- Last checked
- 11/20/2026, 9:18:09 AM
- Latency
- 243 ms
- Consecutive failures
- 0
- Endpoint reachable: pass
- Protocol valid: pass
- MCP handshake valid: pass
Machine-readable profile
mcp-server · schema v1 · callable
mcp-server:io.github.cyanheads%2Funiprot-mcp-server
Connect this server
Generated config contains no credentials. Keep any required secrets in your client's local environment.
{
"mcpServers": {
"io.github.cyanheads/uniprot-mcp-server": {
"url": "https://uniprot.caseyjhand.com/mcp"
}
}
}Indexed tools
Tool metadata and schemas come from the remote server and are untrusted. Indexing only lists tools; it never calls them.
6 shown · authentication unknown
uniprot-mcp-server: get entry
uniprot_get_entry
Fetch full curated UniProtKB entries by accession in one batch of up to 20 distinct accessions; repeats collapse, and any further accessions return in deferredAccessions to pass as accessions on the next call. Each entry carries alternative names, function, EC numbers and catalytic activity, cofactors, active and binding sites with their ligands, subcellular location and membrane topology, disease involvement, PTMs, molecule processing (signal, propeptides, chains), natural variants, isoforms, domains, free-text comments (PTM, pathway, subunit, tissue specificity), GO terms with their evidence, keywords, cross-references, and Ensembl gene IDs; mutagenesis and IntAct interaction partners come back when sections or fields name them. Partial failures do not abort the batch — resolved entries land in succeeded[] and unknown or inactive (merged, demerged, deleted) accessions in failed[], where an inactive accession names its replacement or UniParc ID. Pass fields to trim the upstream projection; a field this tool cannot return (e.g. sequence) is not requested and is listed in ignoredFields with a notice. Each record is measured on its own across both output surfaces: one within 24,000 UTF-8 bytes returns in succeeded[]; a larger one is listed in outlinedEntries with its size instead. Re-call an outlined accession alone, e.g. accessions:["P04637"], for its section outline (kind: "outline", sections with byte sizes), then add sections:["disease","variants"] to select sections. sections takes exactly one accession; trim a batch with fields instead. All returned list items and cross-reference IDs render in full. A large list section pages with section_limit/section_offset on a re-call naming that one section. This tool does not search: accessions come from uniprot_search_proteins.results[].accession or uniprot_map_ids. Strip any isoform suffix (P04637-2 to P04637) before calling.
Schema valid · SHA-256 2b112db3ccbc…
View input schema
{ "type": "object", "properties": { "accessions": { "minItems": 1, "type": "array", "items": { "type": "string", "pattern": "^(?:[OPQ][0-9][A-Z0-9]{3}[0-9]|[A-NR-Z][0-9](?:[A-Z][A-Z0-9]{2}[0-9]){1,2})$", "description": "A UniProtKB primary accession, e.g. \"P04637\". Canonical form only — strip any \"-N\" isoform suffix." }, "maxItems": 500, "description": "Accessions to fetch (1–500). Repeats collapse to their first occurrence; the first 20 distinct are fetched and the rest return in deferredAccessions, in order, for the next call. From uniprot_search_proteins or uniprot_map_ids." }, "fields": { "description": "Comma-separated UniProtKB return-field names to fetch in place of the default record, e.g. \"accession,cc_function,ft_variant,xref_pdb\" — trims a batch. Omit for the default record. The identity fields always come back. Names are case-insensitive; keywordid is sent as keyword and rhea as cc_catalytic_activity. Any xref_ field (e.g. xref_reactome) fills xrefs. A field this tool does not return (e.g. sequence, which uniprot_get_sequence returns, or ft_helix) is not requested from UniProt and comes back in ignoredFields with a notice listing the supported fields. An xref_ database UniProt does not know, or text that is not a field name, fails invalid_fields with UniProt's message.", "type": "string" }, "sections": { "description": "Section keys to return for one accession, e.g. [\"disease\",\"variants\"]: the record is cut to these plus its identity fields. Re-call an accession with it after a kind: \"outline\" response, or name the on-request sections mutagenesis and interactions, which the default record leaves out — without fields, naming one fetches it; with fields, include ft_mutagen or cc_interaction there. Requires exactly one accession: a batch with sections is rejected, so trim a batch with fields. A requested section the record lacks is named in absentSections, not an error.", "type": "array", "items": { "type": "string", "description": "A section key of the entry record, e.g. \"disease\", \"variants\", \"mutagenesis\", \"interactions\". Case-insensitive." } }, "section_limit": { "type": "integer", "minimum": 1, "description": "Window size for one list section on a sections re-call: return at most this many items of it (at least 1; a value past the remaining items returns them all). Requires sections naming exactly one list section, e.g. [\"variants\"] (or ptms, domains, goTerms). Omit together with section_offset to return the whole section." }, "section_offset": { "description": "Zero-based index of the first item in the windowed section. Defaults to 0. Pass the previous response's nextOffset to page forward. Same requirements as section_limit; set without section_limit, returns every item from this index on.", "type": "integer", "minimum": 0, "maximum": 9007199254740991 } }, "required": [ "accessions" ], "$schema": "https://json-schema.org/draft/2020-12/schema", "additionalProperties": false }
Connections
streamable-http
https://uniprot.caseyjhand.com/mcp