Extensions · permissions · data handling · enterprise distribution
Charles Trinh — Browser Extensions for Clinical Workflows

Extensions that sit in a clinician's browser: permission design, data handling, update channels, and the review that healthcare procurement will actually apply.
What this is
A browser extension runs inside the page where patient data already is, which is exactly why it is useful and exactly why it is scrutinised. Both facts are permanent.
The failure modes are well known and still common: broad host permissions because they were easier, a silent background update, and selectors that break the week the EHR ships a redesign.
This page covers the design decisions that survive a security review and a bad Tuesday: least permission, honest data flow, a controlled update channel and a failure that tells the clinician something is wrong.
Permissions are the product
Request the narrowest host and API permissions; every extra one is a procurement objection and a real risk.
List each permission with the feature that needs it and the feature that would break without it. Permissions nobody can justify are the first thing a reviewer removes, usually along with the tool.
Data handling
State what leaves the browser, where it goes and what is stored locally, and make it enforceable rather than described.
The description should be generated from the code's egress paths, not written independently of them. If the two disagree, procurement will find out before you do.
Update channel
An extension updates itself; pin, sign and control the channel or you have shipped unreviewed code to clinicians.
Self-hosted distribution with explicit versions lets an organisation stage an update and roll back. A public store auto-update is not a controlled change process.
Enterprise distribution
Private listing and policy enforcement beat a public store listing for any regulated deployment.
Private distribution also removes the impersonation risk that a public listing invites, and gives IT a manageable install surface.
DOM fragility
Clinical systems change without notice; selectors break and the extension fails silently in front of a patient.
Prefer stable, accessible attributes over visual position, add a self-check on load, and show a clear warning when the expected structure is missing rather than rendering nothing.
Offline and error behaviour
Decide what the extension does when the API is unreachable, and never block the clinician's underlying workflow.
The host page must work exactly as it did before the extension arrived. An extension that degrades a clinical system during its own outage is a liability with a logo.
Audit
Log which user, which patient context and which action, because the extension is now part of the record path.
The audit needs to be readable by a compliance officer without a developer present, which usually means a plain statement of user, context, action and time.
Removal
A documented uninstall path, or the extension becomes permanent infrastructure nobody owns.
Removal should leave no local data behind and no dangling access grants, and it should be tested as a scenario rather than assumed.
How I do it
I start from the permission request, because it constrains everything else. Each host permission is traced to a feature and each API to a call, and anything that cannot justify itself is removed before the design is final.
The extension is then built to fail visibly: a load-time self-check on the host page structure, an explicit offline state, and an audit record written at the point of action. Local storage is kept to what the user needs to see, not to what is convenient to cache.
Delivery is a private, versioned channel with a tested rollback and a documented uninstall. We run the uninstall once and check nothing is left behind, because that is the scenario a reviewer asks about and the one nobody rehearses.
Artifacts you keep
- A permission table naming the feature each permission serves
- An egress map of every endpoint the extension calls and what it sends
- A private, versioned distribution channel with a tested rollback
- A load-time self-check that warns when the host page structure changes
- An uninstall test confirming local data and grants are cleared
What an extension can and cannot see
An extension runs inside the user's browser with the user's session, which makes it powerful and dangerous in the same breath. It can read what is rendered and act as the user; it cannot be trusted with a broader credential than the session already holds.
Design for the minimum: the specific origins, the specific fields, and no background access to the rest of the web. Broad host permissions are the defect reviewers find first, and they are usually added for convenience.
Update and distribution
An extension is software distributed to browsers, and the update path is the attack path. Sign the package, pin the update source, and document how a forced rollback would work if a release is bad.
Keep the code reviewable and small. An extension with a large dependency tree is an extension whose behaviour you cannot describe, and browser stores increasingly require that description.
Data handling inside the page
Anything captured from a page is PHI if the page is clinical. Send it to the estate's own endpoint, never to a third party, and never store it in extension local storage beyond the session.
Test what is left behind: uninstall the extension and inspect the profile. A surprising amount of clinical text survives in storage, and the user has no way to know.
The honest failure mode
When the target page changes, the extension breaks. Make the break visible and loud rather than silent, and version the selectors so the break is diagnosable in a day.
A silent break in an extension looks like a user who stopped working, which is the failure that reaches the clinic before it reaches the team.
R
eview before distribution
A
n extension is reviewed by whoever installs it and audited by whoever inherits it. Keep the manifest readable, name the permissions and the reason for each, and put the data-handling statement where an administrator will find it rather than in a release note.
W
here the organisation distributes internally, sign the package and keep the signing key with the same discipline as any other production credential.
The support surface
Extensions fail on the user's machine, which means support needs a way to see state without remote access. A diagnostics export produced by the extension itself, containing versions and selector outcomes but no clinical text, is the difference between a resolved ticket and a remote session.
Ship the export from the first release, because it is almost impossible to add to a field install.
Record the extension's version in every payload it sends. When a defect appears in the field, the first question is which build the user is running, and the answer should not require a support call.
Publish the permission list in the internal catalogue with the reason for each permission. The reviewer who inherits the extension reads the catalogue entry before the manifest, and the entry is where a bad permission gets caught.
Review the extension against the browser store policies before each submission, since the policies change and a rejection costs a release cycle.
Keep the policy checklist with the release notes.
Proof
- ONC (g)(10) self-test PASS, 317 of 317 subtests, April 2026 — public suite URL
- 67 public hostnames, 51 serving, one AWS account
- 104 clearinghouse enrolments across 53 jurisdictions
- USCDI v7 proposal — Patient Digital Identity Credential — listed at isp.healthit.gov
- SAM.gov UEI MVKQW9DKKGB8 · CAGE 20KM5 · NPI 1194645564
- AWS Business Associate Addendum active since 18 August 2026
Where this work is delivered
Engagements run from Las Vegas and reach across the valley: clients in Las Vegas, NV; Las Vegas, Nevada; Henderson; Clark County, Nevada; and North Las Vegas, Nevada, plus remote delivery nationwide.
How an engagement runs
1. Trace one real case
Follow a single patient or claim end to end with a timestamp at every hop, before any document is written.
2. Written findings
A findings list naming the defect, the system, the owner and the date.
3. Monthly review
Review the list monthly and check whether the dates held.
Frequently asked
What is this service?
This page describes charles trinh — browser extensions for clinical workflows — the scope, the failure modes it prevents and how the work is measured. It sits in the Security and Compliance category at trinh.io.
Where is the work delivered?
Engagements run from Las Vegas, Nevada across the valley — Las Vegas, Henderson, North Las Vegas, Spring Valley, Paradise, Winchester, Sunrise Manor, Enterprise, Summerlin, Clark County — and remotely anywhere in the United States. On-site work in the Las Vegas valley, remote delivery elsewhere.
What does it cost?
Standard rate $195 per hour, committed $165 per hour at twenty hours or more per week for three months or more; $1,600 standard day, $1,350 committed day. Scoped in writing before work starts, fixed fee where the deliverable is defined.
Who is Charles Trinh?
A technology architect in Las Vegas, Nevada, working in healthcare identity, consent and interoperability. Not a physician, not a medical practice, and not affiliated with any other Charles Trinh.
How does an engagement start?
With one trace of a real case and a written findings list naming the defect, the system, the owner and the date. The first call is twenty minutes and free: 702-559-3489.
Engagement
Scoped in writing before work starts; fixed fee where the deliverable is defined. First call is 20 minutes and free. 702-559-3489 · charles@trinh.io
Talk to Charles
- Phone: 702-559-3489
- Email: charles@trinh.io
- Based in Las Vegas, Nevada · working remotely across the United States