Skip to main content

AI Skills Catalog

AI Skills for care facility owners: reusable Claude workflows you can run this week

An AI Skill is a repository definition with a prompt, allowed tools, and an output contract. This page presents a curated public subset of thirteen definitions covering inbound triage, sales prep, SEO, content, expense categorization, and codebase QA. Status varies by definition; presence does not imply testing, scheduling, PHI approval, or authority to act externally.

What is an AI Skill?

A single-purpose Claude Code function. You give it inputs in a structured format, it runs a fixed prompt against a fixed tool list, and it returns a structured output. There is no chat, no drift, no surprise. Skills are how we make AI work reproducible — and auditable when a state surveyor asks how a document was generated.

The 13 skills

1 of 13 shipped · the rest scheduled Q3 2026

  • Inquiry Classifier

    Planned
    /inquiry-classifier

    6-intent triage of inbound messages — eMAR demo, consulting fit-call, partnership, support, vendor/spam, unclassified.

    Used by:
    AI Secretary
    Cadence:
    Per inbound message

  • Sales Pre-Call Brief

    Planned
    /sales-pre-call-brief

    Per-prospect 1-page dossier 24 hours before a fit call: facility profile, likely pain points, opening 3 minutes.

    Used by:
    AI Sales Engineer
    Cadence:
    Per fit call

  • SEO Audit (GBP)

    Planned
    /seo-audit-gbp

    Weekly Google Business Profile + local SEO audit: NAP consistency, review velocity, schema coverage, AI-search extractability.

    Used by:
    AI SEO Specialist
    Cadence:
    Weekly

  • Feeder Vertical Page Builder

    Planned
    /feeder-vertical-page-builder

    Per-vertical landing page generator with answer-first H2 structure, FAQ section, and full Service + FAQPage + BreadcrumbList schema.

    Used by:
    AI SEO Specialist, AI Content Writer
    Cadence:
    Per new vertical

  • Social Post Generator

    Planned
    /social-post-generator

    Weekly batch of LinkedIn, X, and Instagram posts from one theme — platform-tailored hooks, lengths, hashtags, and CTAs.

    Used by:
    AI Social Manager
    Cadence:
    Weekly

  • CMO Brief (Weekly)

    Planned
    /cmo-brief-week

    Friday synthesis: pipeline health, content pillar performance, channel signal, the 1-3 decisions the founder must make next week.

    Used by:
    AI CMO
    Cadence:
    Friday weekly

  • Founder Brief

    Planned
    /founder-brief

    Weekly synthesis of every other AI Employee's output + decision log, written for a founder to read in 5 minutes.

    Used by:
    AI Chief of Staff
    Cadence:
    Sunday weekly

  • Partnership Email Draft

    Planned
    /partnership-email-draft

    1:1 personalized outreach to institutional partners — bar associations, CLE providers, state associations, referral networks.

    Used by:
    AI Head of Partnerships
    Cadence:
    Per target

  • Creative Image Brief

    Planned
    /creative-image-brief

    Multi-model image prompts (DALL-E, Midjourney, Imagen) with brand-consistent style guides for blog and social.

    Used by:
    AI Head of Creative
    Cadence:
    Per asset

  • Dynamic Pitch Deck Builder

    Planned
    /dynamic-pitch-deck-builder

    Per-vertical sales deck (A1/A2/A3/B1/B3/B4 tracks) assembled from a slide library with vertical-specific stats and quotes.

    Used by:
    AI Sales Engineer
    Cadence:
    Per prospect

  • Expense Categorizer

    Planned
    /expense-categorizer

    Transaction categorization against the Chart of Accounts; flags ambiguous entries for human review.

    Used by:
    AI Accountant
    Cadence:
    Monthly

  • Data Pulse

    Planned
    /data-pulse

    Weekly multi-source KPI roll-up: LTV/CAC, MRR, churn, pipeline velocity, content pillar engagement.

    Used by:
    AI CMO, AI Head of COO
    Cadence:
    Weekly

  • React Doctor

    Shipped
    npx react-doctor

    React codebase health score: lint, dead code, accessibility, bundle size, architecture diagnostics. Catches regressions before commit.

    Used by:
    AI Build QA Reviewer
    Cadence:
    Per commit

Frequently asked questions

What is an AI Skill?
An AI Skill is a reusable Claude Code slash command: a deterministic prompt + a tool list (file read, file write, web fetch, bash) + an output contract that specifies exactly what the skill returns. Skills are designed to be run from the terminal in seconds, invoked by other Claude Code sessions, or chained inside AI Employees. They are the smallest unit of automation in the CareMAR catalog.
How is an AI Skill different from a chatbot?
A chatbot is a free-form conversation; an AI Skill is a one-shot function. You invoke a skill with a structured input, it does deterministic work using a fixed tool list, and it returns a structured output. There is no chat history, no drift between runs, and no surprise behavior — the output contract is part of the spec. Skills are how we keep AI work auditable and reproducible.
Can I run these without writing code?
Shipped skills include documented invocation and output instructions. Setup effort depends on your environment and access; the walkthrough covers installation and a first supervised run.
Do they work with HIPAA data?
The currently shipped business-operations skills are not approved for PHI. Any future workflow that could handle resident data requires a documented data-flow review, appropriate vendor and BAA confirmation, and facility approval before use.
How do I pick which skill to start with?
Start with the one closest to a task you already do every week and dread. If you write social posts, start with Social Post Generator. If you do monthly bookkeeping, Expense Categorizer. If you take sales calls, Sales Pre-Call Brief. The point of an AI Skill is to remove the recurring cognitive load — the highest-value pilot is whichever task is currently on your calendar this week.
What does “Planned” status mean?
Twelve of the thirteen skills are documented with their full prompt, tool list, output contract, and target queries, but the .claude/skills/<slug>/SKILL.md file is not yet committed to the public repo. Planned skills ship in Q3 2026. React Doctor is the one currently shipped — install it with npx react-doctor and run it against any React project today.

Want help picking your first skill?

Thirty minutes on a call. We identify one reviewable, non-PHI task and show which definition could prepare a supervised first draft.

Book a discovery call