If you use Claude in your clinical or administrative work — writing referral letters, thinking through differentials, staying current on guidelines — you've noticed the same limitation every session: Claude starts fresh. It doesn't know your specialty, your prescribing preferences, your usual patient population, or the protocols your practice follows.
You end up re-explaining the same context every time. That's not how a good assistant should work.
The context that makes Claude useful isn't patient records — it's the professional frame you carry in your head:
None of this is PHI. All of it saves you from re-briefing Claude on who you are and how you work.
You: "I'm a GP in the UK. I want to think through a presentation I saw today — 68-year-old with new-onset dyspnoea and bilateral ankle oedema. Help me think through the differential."
Claude: "Happy to help. To give you the most relevant thinking, could you tell me your clinical context — are you in primary care, secondary care...?"
You re-explain your specialty, location, guidelines, and context before getting to the actual question.
You: "context()"
Claude: [reads your Stash context — UK GP, inner-city practice, NICE guidelines, multimorbid patient population]
You: "68-year-old presenting with new-onset dyspnoea and bilateral ankle oedema. Walk me through the differential."
Claude: "Given your primary care setting and patient population, the differential starts with heart failure — BNF dose thresholds for diuretics you'd consider at initiation are... For NICE CG108 referral criteria, the threshold is..."
Contextually accurate from the first response. No re-briefing.
Beyond clinical context, doctors use Stash as a personal knowledge log for their CPD and research work:
These are personal professional notes. They're not patient records. They're the kind of thing you'd write in a paper notebook — except Claude can search and reason across them.
One underrated use: storing your administrative preferences and templates so Claude drafts letters that match your house style from the first sentence.
Example Stash records for an outpatient physician:
Not clinical records. Not PHI. Just the institutional knowledge that makes AI-assisted drafting actually useful.
The context() tool returns your stored professional context at the start of a session. Some doctors use a variation of the "Start my day" pattern to open a focused work session:
You: "context(expand=['research-queue', 'cpd-log'])"
Claude: [loads your professional profile + outstanding research questions + recent CPD entries]
You: "I have 30 minutes before clinic. What are the three oldest items in my research queue, and is there anything urgent in my CPD log I should address before my appraisal next month?"
You get a prioritised briefing drawn from your actual stored context — not Claude's general knowledge about what doctors typically need.
| Not Stash | Why |
|---|---|
| Patient records or EMR | Clinical records need HIPAA/GDPR-certified storage with access controls, audit logs, and data retention policies. Stash has none of that. |
| Clinical decision support | Stash stores your context. Claude reasons across it. But you are still the clinician — the outputs are prompts for your thinking, not clinical decisions. |
| Dictation or SOAP notes | There are purpose-built AI scribing tools for this. Stash is not one of them. |
add a few key facts: your specialty, your prescribing context, guidelines you followcontext() — Claude will have your professional frame loaded without re-briefingFree tier: 2,500 records / 50 searches per month. One collection per specialty, another for CPD, another for research questions — that's well within the free tier.
Add Stash to Claude
Connector URL (paste into Claude Settings → Integrations → Add MCP):
https://app.stashlite.com/mcp
Sign in with Google. No card required.