@pipeworx/symmap

SymMap v2 (symmap.org, Beijing University of Chinese Medicine) — the Traditional Chinese Medicine association graph: 698 herbs, ~26k ingredients, ~21k protein targets, TCM symptoms, TCM syndromes, modern medical symptoms and ~14k diseases, connected by curated and predicted associations. Every relationship payload carries an explicit evidence_tier (traditional_use | human_clinical | laboratory | computational_prediction) and an evidence_basis sentence, because SymMap records associations, not efficacy — nothing in it is clinical proof that a herb treats anything.

Part of Pipeworx — an MCP gateway connecting AI agents to 1546+ live data sources.

Tools

• symmap_search(query, entity) — resolve a name (Chinese, pinyin, Latin, English, gene symbol, disease name) to SymMap ids across any of the seven entity classes.

• symmap_herb(herb) — full bilingual herb record: TCM properties, meridians, drug class, used part, all name forms.

• symmap_herb_ingredients(herb) — molecules identified in a herb, with PubChem CID, CAS and oral-bioavailability score.

• symmap_herb_symptoms(herb) — the herb's traditional indications: TCM symptoms (SMTS) and TCM syndromes (SMSY), kept separate from the modern vocabulary.

• symmap_symptom_herbs(symptom, system) — herbs for a symptom; system:"tcm" is a direct lookup, system:"modern" goes through SymMap's curated modern→TCM symptom crosswalk with both hops explicit.

• symmap_ingredient(ingredient) — an ingredient's protein targets (PubMed-cited edges are tier laboratory, uncited ones computational_prediction) and the herbs it occurs in.

• symmap_target_diseases(target) — diseases associated with a gene; OMIM/Orphanet-backed edges are tier human_clinical, the rest computational_prediction.

Related MCP server: dayi-search-mcp

Auth

Keyless.

Data sources

• http://www.symmap.org/related_components/ — form-POST JSON: association edges for one entity (rrid, table_name, filter).

• http://www.symmap.org/search/ — form-POST JSON: entity search by name (table_name, key).

• src/herbs.ts — the herb entity table, generated by scripts/bake-herbs.py from the SMHB file on SymMap's download page, so herb lookup never depends on the upstream server.

Things the next person would rediscover the hard way:

• The site is plain HTTP only — https://www.symmap.org does not answer.

• The downloadable "key files" are NOT relation edges. They are long-format (id, Field_name, Field_context) dumps of the same entity tables. The actual graph edges are only served by /related_components/.

• Some join directions 502 by design of their backend: herb→Gene, herb→Disease, herb→MM_symptom and MM_symptom→Herb all return 502 every time, while herb→Mol, herb→TCM_symptom, herb→Syndrome, Mol→Gene, Mol→Herb, TCM_symptom↔MM_symptom, TCM_symptom→Herb, Syndrome→Herb, Gene→Disease and Disease→Gene answer reliably. The tools compose only the reliable directions; a herb's targets are reached herb→ingredients→targets, one hop at a time.

• rrid accepts both SMHB1 and SMHB00001 forms.

• In v2.0, SMTS is TCM symptoms, SMSY is TCM syndromes (new in v2), SMMS is modern-medicine symptoms. The v1 downloads reuse SMTS differently; don't mix versions.

• Ingredient→target literature evidence arrives as an HTML snippet with the same sentence twice (short + long variants) plus a Fold button; the pack extracts the long variant and the PMIDs.

Licence / attribution

SymMap is a free-to-access academic database; its site states the download files are free to download and carries no further reuse terms. Cite the paper when publishing results: Wu et al., SymMap: an integrative database of traditional Chinese medicine enhanced by symptom mapping, Nucleic Acids Research 2019, doi:10.1093/nar/gky1021.

Quick Start

Add to your MCP client (Claude Desktop, Cursor, Windsurf, etc.):

{
  "mcpServers": {
    "symmap": {
      "url": "https://gateway.pipeworx.io/symmap/mcp"
    }
  }
}

What this endpoint actually serves

tools/list at https://gateway.pipeworx.io/symmap/mcp returns the tools in the table above plus the shared Pipeworx meta-tools — ask_pipeworx, discover_tools, search_within, remember/recall and the rest of the gateway-wide set. So the tool count you see is larger than this table: a single-pack endpoint currently lists roughly 30 shared tools alongside the pack's own. The connection's initialize response states its exact scope, and is the authoritative answer for a given day.

This is deliberate, not multiplexing by accident. The meta-tools are what let a scoped connection answer a question this pack does not cover — via ask_pipeworx, which routes across the whole catalog — without you adding a second MCP server. There is currently no way to mount a pack endpoint without them; if the extra schemas cost you more context than the routing is worth, connect to the full gateway once rather than to several pack endpoints.

Or connect to the full Pipeworx gateway to get every pack's tools listed directly, instead of just this one's:

{
  "mcpServers": {
    "pipeworx": {
      "url": "https://gateway.pipeworx.io/mcp"
    }
  }
}

Both URLs reach the same gateway and the same 1546+ data sources. The only difference is which pack's tools are listed directly; ask_pipeworx reaches all of them from either one.

Using with ask_pipeworx

Instead of calling tools directly, you can ask questions in plain English — this works on the pack endpoint above as well as on the full gateway:

ask_pipeworx({ question: "your question about Symmap data" })

The gateway picks the right tool and fills the arguments automatically.

More

• Docs and guides

• pipeworx.io

License

MIT