Installation

Install HealthChain with pip:

pip install healthchain

The core install is deliberately lean — type-safe FHIR, gateways, the agent toolkit, and the CLI, with no ML or NLP frameworks attached. Optional capabilities ship as extras:

Extra	Installs	When you need it
`cda`	`python-liquid`, `xmltodict`	CDA parsing and NoteReader integration (`CdaAdapter`, interop engine)
`mcp`	`mcp`	Serving the `FHIRToolkit` over MCP (`kit.as_mcp()`, `healthchain mcp`)
`langchain`	`langchain-core`	`FHIRToolkit.as_langchain()` for LangChain agents
`sandbox`	`faker`	Synthetic FHIR test data generators for the sandbox client
`examples`	`spacy`, `transformers`	Running the NLP/ML cookbook examples

pip install "healthchain[cda,mcp]"   # combine as needed

Bring your own model framework and use it directly in pipeline nodes. Each cookbook states the exact install line it needs, including any model-specific dependencies (e.g. torch).

Give your AI assistant the API

Claude Code users can install the HealthChain plugin, which teaches Claude the current API, the CLI verbs, and the recipes:

/plugin marketplace add healthchainai/HealthChain
/plugin install healthchain@healthchain

Codex users can install the same skill from the repository marketplace:

codex plugin marketplace add healthchainai/HealthChain
codex plugin add healthchain@healthchain

Any other assistant can read llms.txt for a map of the API docs.