Claude Code for healthcare — the engineer-grade Claude consultancy

What Claude Code is

A CLI agent for engineers. Installed with one npm command. It can read your repo, write to your repo, run bash, fetch the web, and invoke any slash command defined in .claude/skills/. It is the opposite of a chatbot: deterministic, scriptable, and version- controllable. Engineers treat it as the agentic layer on top of their existing terminal workflow.

How CareMAR uses it

Every AI Skill in the catalog is a Claude Code slash command, committed to the public repo under .claude/skills/. Every internal workflow — building this site, maintaining the eMAR product, drafting the weekly CMO brief — runs through Claude Code. The repo's root CLAUDE.md defines the conventions; the .claude/agents/ directory defines the subagents that AI Employees can dispatch to.

When we ship a Claude Code engagement to a customer, the deliverable is the same thing we use ourselves: a configured CLAUDE.md, a set of .claude/skills/ slash commands, and a .claude/agents/ subagent library. The customer can run them, modify them, version them, and extend them. They own the configuration outright.

Frequently asked questions

What is Claude Code?

Claude Code is Anthropic's terminal-based agent for software engineers. It runs in your CLI; it can read and write files in your repo, run bash, invoke custom slash commands, and chain subagents. It is the runtime for AI Skills — the deterministic, reusable Claude functions in the CareMAR catalog. You install it once and from then on every documented skill is a single slash command away.

How is Claude Code different from Claude Cowork?

Cowork is the browser UI for non-engineers running persistent projects (where AI Employees live). Claude Code is the CLI for engineers running deterministic skills (where AI Skills live). The CareMAR catalog uses both: employees in Cowork, skills in Claude Code, automations that chain across the two. You do not have to pick — they are complementary.

Why hire a Claude Code consultant?

Same reason you hire an AWS consultant: the tool is general-purpose and you want a specific outcome. A Claude Code consultant authors the skills, subagents, and project-level CLAUDE.md files that turn the CLI into a working agent for your business. CareMAR specializes in healthcare and small operator use cases — every skill in the catalog ships with HIPAA scoping notes and is reviewable by the Legal Compliance employee.

How does CareMAR use Claude Code internally?

Every line of code in this site, the eMAR product, the Skills library, and the internal automation scripts is authored with Claude Code in the loop. The skills are versioned in .claude/skills/<slug>/SKILL.md; the subagents are defined in .claude/agents/<slug>.md; the project conventions live in CLAUDE.md at the repo root. We dogfood our own consulting deliverable every day.

Do I need to know how to code to benefit?

To author new skills, yes. To run shipped skills, no — you install Claude Code once (npm install -g @anthropic-ai/claude-code) and from then on every documented skill is a single slash command. The seminars walk you through the install, the first skill run, and how to call a CareMAR-authored skill from your own repo within 30 minutes.

Is Claude Code HIPAA-safe for healthcare use?

Yes when scoped correctly. Claude Code uses the Claude API, which is covered under Anthropic's Enterprise BAA for accounts that need it. We design skills so PHI either stays in your own AWS account (the skill reads metadata, not records) or is excluded from the API call entirely. The Legal Compliance employee reviews any new skill that approaches resident-adjacent data before it ships.