Skip to main content
Glama
stabgan

mcp-icd10

by stabgan
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Local

mcp-icd10

Offline MCP server for ICD medical code lookup, search, and crosswalk translation.

Zero API keys. Zero network calls. Zero data leakage.

124K codes across three coding systems with 102K bidirectional crosswalk mappings, all running locally in a SQLite database.

License: MIT Python 3.10+ MCP

Why?

Existing medical terminology services are thin wrappers around remote APIs — rate-limited, network-dependent, and they send your clinical queries to third-party servers. That's a non-starter for HIPAA-conscious workflows.

This server embeds everything locally. A pre-built SQLite database with FTS5 full-text search ships inside the package. Queries resolve in microseconds. Nothing leaves your machine.

Related MCP server: Medical Terminologies MCP

What's included

System

Codes

Source

ICD-10-CM FY2026

98,186

CMS.gov (incl. 74,719 billable)

ICD-9-CM

14,567

CMS GEMs package

ICD-10 WHO 2019

11,243

WHO

GEMs crosswalk

102,591

CMS (bidirectional ICD-9 ↔ ICD-10)

Total: 123,996 codes + 102,591 crosswalk mappings

Performance

All benchmarks on Apple Silicon. Every query hits an index or FTS5 — no table scans.

Operation

Latency

Exact code lookup

4.4 µs

Full-text search

0.23 ms

Category browse

0.04 ms

Crosswalk translation

5.9 µs

DB decompress (first run only)

~38 ms

Quick start

Claude Desktop / Kiro / any MCP client

{
  "mcpServers": {
    "icd": {
      "command": "uvx",
      "args": ["mcp-icd10"]
    }
  }
}

Docker

docker build -t mcp-icd10 .
{
  "mcpServers": {
    "icd": {
      "command": "docker",
      "args": ["run", "--rm", "-i", "mcp-icd10"]
    }
  }
}

Install from source

git clone https://github.com/stabgan/mcp-icd10.git
cd mcp-icd10
pip install .
mcp-icd10

Tools

lookup_code

Look up a medical code and return its description. Accepts codes with or without dots. Searches across all systems when no system is specified.

> lookup_code("E11.9")
[icd10cm] E119: Type 2 diabetes mellitus without complications

> lookup_code("250.00")
[icd9cm] 25000: Diabetes mellitus without mention of complication, type II or unspecified type, not stated as uncontrolled

search_codes

Full-text search across all 124K code descriptions. Supports natural language clinical queries.

> search_codes("acute myocardial infarction")
Found 20 result(s) for 'acute myocardial infarction':
  [icd10cm] I219: Acute myocardial infarction, unspecified
  [icd10cm] I2101: ST elevation (STEMI) myocardial infarction involving left main coronary artery
  [icd9cm] 41071: Acute myocardial infarction of subendocardial wall, initial episode of care
  ...

browse_category

Browse all codes under a category prefix within a specific system.

> browse_category("E11", system="icd10cm")
50 code(s) under 'E11' [icd10cm]:
  E1100: Type 2 diabetes mellitus with hyperosmolarity without nonketotic hyperglycemic-hyperosmolar coma
  E1101: Type 2 diabetes mellitus with hyperosmolarity with coma
  ...

translate_code

Translate between ICD-9-CM and ICD-10-CM using the CMS General Equivalence Mappings (GEMs).

> translate_code("250.00")
Crosswalk for '25000':
  25000 → [icd10cm] E119 — Type 2 diabetes mellitus without complications [≈]

> translate_code("E119", source_system="icd10cm")
Crosswalk for 'E119':
  E119 → [icd9cm] 25000 — Diabetes mellitus without mention of complication, type II ... [≈]

get_stats

Returns code counts per system and total crosswalk mappings.

Architecture

┌─────────────────────────────────────────────┐
│         Claude / Kiro / AI Agent            │
│              (MCP Client)                   │
└──────────────────┬──────────────────────────┘
                   │ MCP Protocol (stdio)
┌──────────────────▼──────────────────────────┐
│            mcp-icd10 Server                 │
│  ┌────────────────────────────────────────┐ │
│  │  FastMCP — 5 tools                     │ │
│  │  lookup · search · browse              │ │
│  │  translate · stats                     │ │
│  └──────────────┬─────────────────────────┘ │
│  ┌──────────────▼─────────────────────────┐ │
│  │  SQLite + FTS5 (read-only)             │ │
│  │  124K codes │ 102K crosswalk mappings  │ │
│  │  Integer system IDs │ Bitmask flags    │ │
│  └──────────────┬─────────────────────────┘ │
│  ┌──────────────▼─────────────────────────┐ │
│  │  Pre-built DB (5.2 MB gzipped)         │ │
│  │  Decompresses on first run → 23 MB     │ │
│  └────────────────────────────────────────┘ │
└─────────────────────────────────────────────┘
           100% local · zero network calls

How it works

The package ships a pre-built, gzipped SQLite database (icd.db.gz, ~5 MB). On first run, it decompresses to ~23 MB and opens in read-only mode with memory-mapped I/O.

Key design decisions:

  • Integer system IDs (0/1/2) instead of text strings for compact storage and fast comparisons

  • CROSS JOIN in FTS queries forces SQLite's query planner to use the FTS index first, avoiding full table scans on broad search terms

  • No FTS5 ranking — ORDER BY rank forces full-scan scoring on all matches. Instead, we cap at 500 FTS candidates and sort by description length (shorter = more specific = better relevance proxy for medical codes)

  • Bitmask flags on crosswalk entries pack approximate/no-map/combination into a single integer

  • WITHOUT ROWID on the crosswalk table for clustered primary key access

Data sources

Limitations

  • No ICD-10-PCS (procedure codes), ICD-11, or SNOMED CT

  • GEMs crosswalk covers ICD-9-CM ↔ ICD-10-CM only (not WHO codes)

  • FTS5 uses Porter stemming — some medical abbreviations may not stem perfectly

  • Update frequency depends on new package releases

License

MIT

Available Tools

5 tools
browse_categoryA

Browse all codes under a category prefix.

Args: prefix: Category prefix (e.g., 'E11', 'I25', '250'). system: 'icd10cm' (default), 'icd9cm', or 'icd10who'. limit: Max results (default 50, max 100).

ParametersJSON Schema
NameRequiredDescriptionDefault
limitNo
prefixYes
systemNoicd10cm

Output Schema

ParametersJSON Schema
NameRequiredDescription
resultYes

TDQS

A4.2/5.0
Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

With no annotations, the description bears full responsibility for behavioral traits. It discloses defaults (limit 50, system icd10cm) and maximum limit (100), which is helpful, but does not explicitly state read-only behavior or side effects. The tool name 'browse' implies read-only, but not explicit.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is concise and well-structured: a single-sentence purpose followed by clearly formatted parameter bullet points. Every sentence is useful, and the most critical information (purpose, prefix, system, limit) is front-loaded.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the presence of an output schema (not shown), the description need not explain return values. It covers the core purpose, all parameters with defaults and constraints, and differentiates from siblings implicitly. The description is complete for a straightforward browsing tool.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters5/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 0% (no property descriptions in schema), so the description must compensate. It provides detailed semantics: prefix examples ('E11', 'I25'), system values ('icd10cm', 'icd9cm', 'icd10who'), limit default (50) and max (100). This adds significant value beyond the schema.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool's purpose: 'Browse all codes under a category prefix.' The verb 'browse' and resource 'codes under a category prefix' are specific and distinct from sibling tools like lookup_code (single code) and search_codes (query), making it easy for an AI agent to select correctly.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description implies usage through parameter explanations but does not explicitly state when to use this tool over alternatives. No when-not or where-else guidance is provided, so the agent must infer context.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

get_statsA

Database statistics: code counts per system and crosswalk mappings.

ParametersJSON Schema
NameRequiredDescriptionDefault

No parameters

Output Schema

ParametersJSON Schema
NameRequiredDescription
resultYes

TDQS

A3.5/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations provided. Description gives high-level overview but lacks detail on output structure, performance, or data freshness. Output schema exists but not described.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Single sentence, front-loaded, no wasted words. Efficiently communicates purpose.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Adequate for a parameterless tool with output schema, but could be improved by detailing output structure or providing example. Vague on 'crosswalk mappings'.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

No parameters (0 params = baseline 4). Description adds meaning by specifying the content of statistics beyond the empty schema.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

Description clearly states it returns database statistics: code counts per system and crosswalk mappings. This distinguishes it from siblings like browse_category, lookup_code, search_codes, translate_code which deal with individual codes or searches.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

No guidance on when to use this tool vs alternatives. Does not mention when to prefer get_stats over search_codes or browse_category.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

lookup_codeA

Look up a medical code and return its description.

Args: code: Medical code (e.g., 'E11.9', '250.00'). Dots optional. system: 'icd10cm', 'icd9cm', or 'icd10who'. Empty = search all.

ParametersJSON Schema
NameRequiredDescriptionDefault
codeYes
systemNo

Output Schema

ParametersJSON Schema
NameRequiredDescription
resultYes

TDQS

A4/5.0
Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations are provided, so the description should disclose behavioral traits. It mentions the function returns a description but does not cover error handling, rate limits, or side effects. The tool is simple, but additional transparency would be beneficial.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is extremely concise, using a clear docstring format that front-loads the purpose and efficiently explains each parameter. No unnecessary words.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the tool's low complexity, the presence of an output schema, and the clear parameter descriptions, the description is nearly complete. It lacks information on edge cases like code not found or invalid system, but overall covers the essential requirements.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The input schema has 0% description coverage, but the description adds meaning by explaining the code parameter with examples and dot optionality, and the system parameter with valid values. This compensates for the lack of schema documentation.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool looks up a medical code and returns its description, with examples and parameter explanations that distinguish it from sibling tools like search_codes and translate_code.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description does not explicitly state when to use this tool versus alternatives, but the sibling names imply that lookup_code is for retrieving a description for a single code, while search_codes is for searching.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

search_codesA

Search codes by clinical description using full-text search.

Args: query: Clinical description or keywords (e.g., 'type 2 diabetes'). system: 'icd10cm', 'icd9cm', or 'icd10who'. Empty = all systems. limit: Max results (default 20, max 50).

ParametersJSON Schema
NameRequiredDescriptionDefault
limitNo
queryYes
systemNo

Output Schema

ParametersJSON Schema
NameRequiredDescription
resultYes

TDQS

A3.9/5.0
Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations are provided, so the description carries full burden. It discloses that search is full-text, system defaults to all, and limit defaults to 20 with max 50. However, it does not mention case sensitivity, matching behavior, ordering, or whether the search is partial/fuzzy. This is adequate but not rich.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is concise: one sentence for purpose, followed by a clean parameter list. No redundancy, efficiently front-loaded.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the tool has an output schema, the description does not need to detail return structure. It covers required input semantics. Missing some context like sorting, but overall sufficient for correct invocation.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters5/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema coverage is 0%, but the description adds thorough semantic meaning: explains 'query' with example, lists allowed 'system' values with note that empty means all, and clarifies 'limit' default and max. This compensates fully for the bare schema.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool's function: 'Search codes by clinical description using full-text search.' It specifies the verb 'Search', the resource 'codes', and the method 'full-text search'. This distinguishes it from siblings like 'browse_category' (browse by category) and 'lookup_code' (lookup specific code).

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description lacks explicit guidance on when to use this tool vs alternatives. It does not state when to prefer this over 'browse_category', 'get_stats', 'lookup_code', or 'translate_code'. No 'when-to-use' or 'use-this-over' statements are present.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

translate_codeA

Translate between ICD-9-CM and ICD-10-CM using GEMs crosswalk.

Args: code: Source code (e.g., '250.00' for ICD-9, 'E119' for ICD-10-CM). source_system: 'icd9cm' or 'icd10cm'. Empty = auto-detect.

ParametersJSON Schema
NameRequiredDescriptionDefault
codeYes
source_systemNo

Output Schema

ParametersJSON Schema
NameRequiredDescription
resultYes

TDQS

A4.1/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations are provided, so the description carries the full burden. It mentions using GEMs crosswalk but does not disclose behavioral traits such as read-only nature, error handling on invalid codes, rate limits, or required authentication. It also does not describe what the output contains beyond the existence of an output schema.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The description is extremely concise: one sentence for the main purpose followed by parameter documentation. Every sentence is informative, with no wasted words. The structure is front-loaded with the core action.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

Given the tool's complexity (2 parameters, no nested objects) and the presence of an output schema, the description is fully adequate. It explains the translation method, parameter formats, and auto-detection behavior. There are no gaps in context for an AI agent to invoke the tool correctly.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

The description adds meaning beyond the input schema: it explains the format of 'code' with examples ('250.00' for ICD-9, 'E119' for ICD-10-CM), specifies the allowed values for 'source_system' ('icd9cm' or 'icd10cm'), and clarifies that an empty value triggers auto-detection. Schema coverage is 0%, so the description fully compensates.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

The description clearly states the tool's action ('Translate between ICD-9-CM and ICD-10-CM') and the method ('using GEMs crosswalk'). The verb 'Translate' and specific resource 'ICD-9-CM and ICD-10-CM' distinguish it from sibling tools that browse, get stats, lookup, or search codes.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The description indicates when to use by specifying the purpose (code translation) and notes auto-detection for the source system. While it does not explicitly state when not to use or name alternatives, the context of sibling tools makes the usage clear.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

  1. 5 tool updatesv0.2.0
    • First observedbrowse_category
    • First observedget_stats
    • First observedlookup_code
    • First observedsearch_codes
    • First observedtranslate_code
Deploy Server
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

A4.1/5.0

Scored across 5 tools

Disambiguation5/5

Each tool has a clearly distinct purpose: browsing categories, getting stats, looking up a specific code, searching by description, and translating between systems. No overlap in functionality.

Naming Consistency5/5

All tool names follow a consistent verb_noun pattern in snake_case (e.g., browse_category, lookup_code), making them predictable and easy to understand.

Tool Count5/5

With 5 tools, the set is well-scoped for an ICD-10 lookup and crosswalk tool. It covers essential operations without being bloated or sparse.

Completeness5/5

The tools provide complete coverage for the domain: browsing by category, exact code lookup, full-text search, crosswalk translation, and database statistics. No obvious gaps for the stated purpose.

Maintenance

ActivityInactive
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers