mteam

"The best-run healthcare companies don't have more people. They have better operational intelligence — they see things their competitors can't see, and they act on them faster." — Every healthcare operator who's ever scaled past 5,000 patients

mteam is the operational intelligence platform for healthcare. 19 AI agents that find the revenue you're leaving on the table, the capacity you didn't know you had, the quality gaps you can't see from any single dashboard, and the scaling bottlenecks that will break before you grow.

The problem: Healthcare operations runs on parallel loops — clinical quality, capacity planning, regulatory compliance, revenue integrity, workforce health, patient flow — all happening simultaneously, all interdependent. A clinical decision affects billing which affects compliance which affects quality metrics. No single person, no single dashboard, no single meeting captures the full picture.

mteam gives you the analytical depth of a €1B health system. Revenue forensics, clinical variation analysis, cross-domain pattern detection, referral network mapping, demand intelligence, and theory-of-constraints scaling analysis. Capabilities that only Kaiser Permanente, Cleveland Clinic, and Mayo currently have — in a repo you can clone in 30 seconds.

Inspired by Gary Tan's gstack . Pure SKILL.md files. Works with Claude Code, Codex, Gemini CLI, or any agent that supports the SKILL.md standard. No code, no dependencies, no runtime. Just markdown that gives AI agents healthcare operational expertise.

gstack helps you BUILD faster. mteam helps you SEE what nobody else sees.

What the first 90 days look like

Week 1-2: /morning-brief + /compliance-check establish the baseline. You discover a regulatory deadline you'd missed, a data feed that's been silently failing, and two clinicians whose mandatory training lapsed.

Week 3-4: /revenue-integrity + /pathway-economics reveal the financial picture you've never seen. You're leaving 8-12% of revenue on the table through incomplete coding. Your assessment pathway has 28% margin but your medication review pathway has 37% margin — you should be doing MORE medication reviews.

Week 5-6: /clinical-variation + /clinical-audit show the quality picture. Three clinicians take 50% longer per assessment with no difference in outcomes. Your guideline adherence rate is 72% when it should be 95%.

Week 7-8: /demand-intelligence + /referral-intelligence show the growth picture. Referrals from two GP clusters have declined 30% — nobody noticed. Meanwhile, demand from a new region is growing 20% month-on-month with zero marketing spend.

Week 9-10: /pattern-detection finds the compound insights. Patients assessed within 10 days of referral have a 3x higher treatment completion rate than those who wait 30+ days. That's a CLINICAL OUTCOME metric disguised as a scheduling metric.

Week 11-12: /scale-readiness synthesises everything. The binding constraint isn't clinician supply — it's scheduling infrastructure. You can hire 10 more clinicians but your system breaks at 8,000 patients. Fix that first.

Quick start

Install (30 seconds)

Requirements: Claude Code or any SKILL.md-compatible agent

Step 1: Install on your machine

git clone https://github.com/myceldigital/mteam.git ~/.claude/skills/mteam

Then tell Claude about the available skills by adding to your project's CLAUDE.md:

## mteam
Healthcare operations intelligence. Available skills:
/morning-brief, /ops-plan, /workforce-check, /system-health,
/change-management, /revenue-integrity, /pathway-economics,
/demand-intelligence, /clinical-audit, /clinical-variation,
/pattern-detection, /referral-intelligence, /compliance-check,
/incident-response, /data-quality, /predictive-ops,
/scale-readiness, /performance-report, /deep-scan

Step 2: Configure your jurisdiction

Copy config/ireland.md to config/active.md (or uk-england.md, uk-northern-ireland.md, us-federal.md)

Step 3: Run your first scan

You: /deep-scan
     [attach: semble-appointments-aug2025-mar2026.csv]

Claude runs ALL analyses — no questions asked:
- Capacity intelligence (utilisation, no-shows, forecast)
- Revenue intelligence (per clinician hour, per appointment type, gaps)
- Clinical variation (duration differences, capacity hidden in variation)
- Demand patterns (trends, seasonality, referral shifts)
- Workforce health (burnout risk, workload equity)
- Anomaly detection (outliers, data quality, things that don't look right)

→ TOP 5 INSIGHTS ranked by financial/operational impact

Or start with the daily rhythm:

You: /morning-brief

The 19 agents

🛡️ The Defensive Line — Keep the operation safe

Skill	Agent	What it does
/morning-brief	Ops Lead	Synthesises overnight incidents, today's capacity, provider availability, and regulatory deadlines into a structured briefing
/compliance-check	Governance Lead	Runs jurisdiction-specific regulatory checklists with 30/60/90 day lookahead
/incident-response	Risk Manager	Structured response with time-based escalation — immediate safety through corrective action
/data-quality	Data Analyst	Audits clinical records for completeness, coding accuracy, duplicates, and consent gaps

⚙️ The Engine Room — Keep the machine running

Skill	Agent	What it does
/ops-plan	Capacity Analyst	Forecasts demand vs supply, identifies bottlenecks, models roster scenarios
/workforce-check	People Analyst	Flags utilisation outliers, burnout risk, credential expiry, training gaps
/system-health	Systems Engineer	Monitors EHR, lab, pharmacy pipeline health — uptime, failed messages, stale feeds
/change-management	Implementation Lead	Structures rollout plans with stakeholder mapping, training, adoption metrics

💰 The Revenue Engine — Find the money

Skill	Agent	What it does
/revenue-integrity	Revenue Forensic	Analyses billing leakage — unbilled services, coding gaps, claims rejections, aged debt
/pathway-economics	Unit Economics Analyst	Decomposes cost and revenue to the pathway level — margin by pathway, loss-making services
/demand-intelligence	Market Analyst	Analyses referral trends, geographic demand, seasonal patterns, competitor positioning

🔍 The Intelligence Unit — See what nobody else sees

Skill	Agent	What it does
/clinical-audit	Quality Lead	Compares actual care against guideline recommendations — flags overdue, off-protocol patients
/clinical-variation	Variation Analyst	Identifies unwarranted variation across providers — the highest-leverage quality and cost intervention
/pattern-detection	Cross-Domain Intelligence	Connects datasets that never talked — scheduling × outcomes, provider × satisfaction
/referral-intelligence	Referral Analyst	Maps your referral network — conversion rates, relationship trends, leakage points

🎯 The Strategic Layer — Where are we going

Skill	Agent	What it does
/predictive-ops	Leading Indicator Designer	Identifies early-warning signals that predict problems 2-4 weeks before they become crises
/scale-readiness	Scaling Strategist	Theory-of-constraints analysis — what breaks first when you grow
/performance-report	Executive Analyst	Compiles all domains into structured reports with trend analysis and narrative synthesis

📊 Data-Driven — Feed it data, get insights

Skill	Agent	What it does
/deep-scan	Operational Intelligence Engine	Feed it a CSV (appointment export, referral data, financials) and it runs ALL analyses automatically — capacity, revenue, variation, demand, workforce, anomalies — with zero questions. Produces top 5 ranked insights with €-impact estimates

/deep-scan is the flagship agent. The other 18 agents ask questions. This one analyses data. Give it your Semble appointment export and it tells you what it found — including things you didn't know to ask about.

Architecture

mteam is pure markdown. No code, no dependencies, no runtime.

mteam/
├── README.md
├── CLAUDE.md
├── LICENSE
├── config/
│   ├── active.md           ← your jurisdiction (copy from templates)
│   ├── ireland.md          ← HIQA, Medical Council, DPC, GDPR
│   ├── uk-england.md       ← CQC, GMC, ICO, UK GDPR
│   ├── uk-northern-ireland.md  ← RQIA, GMC, ICO
│   └── us-federal.md       ← CMS, state boards, HIPAA
├── context/
│   └── CONTEXT.md          ← shared operational state (agents read/write)
├── checklists/
│   ├── clinical-safety.md
│   ├── data-protection.md
│   ├── regulatory-compliance.md
│   └── incident-reporting.md
├── [18 agent directories]/
│   └── SKILL.md            ← the agent definition
└── docs/
    ├── agents.md           ← deep dives on each agent
    └── philosophy.md       ← operational philosophy

Key design decisions

Jurisdiction-aware. Each agent reads config/active.md to know which regulatory body, data protection regime, and accreditation framework applies. A /compliance-check in Ireland references HIQA standards; in England, CQC. Same agent, different context.

Shared context. context/CONTEXT.md is read and updated by agents. When /incident-response logs an incident, /morning-brief sees it. When /clinical-audit finds a quality gap, /compliance-check knows about it.

Mandatory safety layer. Every agent ends with: "Before finalising, verify: (1) No clinical safety risk. (2) No regulatory obligation missed. (3) Data handling compliant. (4) If uncertain, state the limitation and recommend expert consultation."

Herald integration. /clinical-audit can reference Herald-parsed guideline JSONs to compare actual care against guideline recommendations.

What makes this different

The first 8 agents are operational hygiene any healthcare company should have. The next 10 are analytical capabilities that only €1B+ health systems currently possess. Revenue forensics. Clinical variation analysis. Cross-domain pattern detection. Referral network mapping. Demand intelligence. Theory-of-constraints scaling analysis.

mteam gives a 6,000-patient clinic the operational intelligence of Kaiser Permanente.

Privacy and safety

• No patient data in any SKILL.md file or in this repo
• Agents help you THINK about operations — they don't process patient records
• mteam helps you structure queries for your EHR, but actual data stays in your clinical systems
• Every agent output passes through a clinical safety and regulatory compliance check
• See REGULATORY.md for classification and regulatory position

License

MIT. Free forever. Go operate something.

---

Built by Matthew Gavin — same pattern as gstack, purpose-built for regulated healthcare.