Packaged for IPM/ZPM: module.xml with Interoperability production, Workflow Engine approval loop and Embedded Python agent.

1.0.5 — Optional Java PEX handoff

• New: an optional Java Business Operation built with InterSystems PEX. Storm.JavaHandoff (EnsLib.PEX.BusinessOperation) runs storm.pex.SafeHandoffOperation through the %Java Server gateway and turns each newly saved root cause into a short on-call handoff card, shown in the storm's own Visual Trace session.
• Off by default. A standard install behaves exactly as 1.0.4; the README section "Java PEX (optional)" shows how to switch it on, what to look for, the recovery step and how to switch it off again: https://github.com/inunanba/storm-link#java-pex-optional
• The FINAL result is returned before Java is called. In fresh isolated IPM installs on IRIS 2026.3 AI Hub build 154 with Java 11, storms A1 and A2 gave the same results and Managed Alerts with Java PEX off and on, and the native suites passed (12/72, Workflow 9/64): https://github.com/inunanba/storm-link/blob/main/docs/java-pex-1.0.5-ipm-run.txt
• Java source and build script are in pex/; the shipped jar rebuilds to the same class files.

MCP servers in Settings, on your own install

• An admin can now add an MCP server in Settings > Tools and MCP servers, with an optional secret kept in the IRIS Wallet, and classify each of its tools as Read-only or Risky. Until a tool is classified, it counts as risky. Only admins add or remove a server or classify its tools, and only from the Settings page.
• Every call of a server's tool goes through OcuTalon's tool gate, and each call that runs writes an audit entry. A risky call waits for a person's approval.
• The approval request shows the tool, its MCP server and each of its inputs on a line of its own, with secrets masked, instead of one line cut at 200 characters. You can answer it in the web chat, on the phone page or in Telegram. In Telegram, a request too long for one message shows only its start, with no Approve button.
• Removing a server cancels its pending approval requests.
• The guide has a worked example, A Jira MCP server, step by step. MCP servers are off on the online demo.

Email from the chat, on your own install

• Ask OcuTalon in the chat to send an email. All email goes out from the one mailbox connected in Settings > Channels > Email.
• The admin who connected the mailbox can have OcuTalon email the mailbox's own address without an approval.
• An email to any other address needs an approval, and only an OcuTalon admin can have OcuTalon email another address. Without the admin role, OcuTalon tells you so and sends nothing.
• No email's recipients are in the trace. The trace does keep the chat's messages as written, so an address you type in the chat is in it. Mail is off on the online demo.
  Less of your mailbox in the trace
• Saving the mailbox in Settings > Channels > Email no longer puts its address and user name in the trace, so a user with the trace role can't read them there.

A faster first load

• Settings, History and Usage now load when you open them. The files a browser downloads before OcuTalon's first page shows went from 991,989 bytes in 1.0.9 to 888,402 bytes.

Upgrades clean up after themselves

• An upgrade now removes the earlier versions' web files that the new version doesn't use, and the install log names each file it removes.
• A browser tab left open across an upgrade reloads itself once when you version doesn't use, and the install log names each file it removes.
• A browser tab left open across an upgrade reloads itself once when you open Settings, History or Usage, instead of failing to open the page.

Measured at this release

• The README's Measured table and its agent-versus-fixed-rules comparison now come from this release's evaluation, with Claude Haiku 5.5 on November 14.
• The agent got 22 of 24 routine attention decisions, 11 of 11 incidents and 11 of 12 held-out cases right. It sent one item that needed you now to the digest instead, and made 0 unsafe outcomes. The fixed rules alone got 6 of 24, 0 of 11 and 2 of 12. See With the agent and with fixed rules only.

Requires the InterSystems IRIS for Health AI Hub Early Access image, build 154 or build 109. Upgrading from 1.0.9 keeps your conversations, reminders, watches, approval requests, settings, memories, documents and Telegram pairing. The online demo still runs 1.0.9.

---

Online demo: https://demo.ocutalon.com · Guide: https://github.com/jbrandtmse/OcuTalon/blob/main/GUIDE.md · Open Exchange: https://openexchange.intersystems.com/package/OcuTalon

The online demo is offline during voting: its GitHub codespace's free hours ran out on 14 Nov. Watch the 43 s video (https://raw.githubusercontent.com/drexthealpha/Knos/main/iris/media/knos-iris.mp4) or run Knos with one command on your machine (iris/README.md). The package is the same as 0.1.14.

The online demo is offline during voting: its GitHub codespace's free hours ran out on 14 Nov. Watch the 43 s video (https://raw.githubusercontent.com/drexthealpha/Knos/main/iris/media/knos-iris.mp4) or run Knos with one command on your machine (iris/README.md).

Shutter 1.4.7 adds a FHIR R4 receiver, proved against InterSystems IRIS for Health's own FHIR server and playable on the online demo as scenario S35.

• A failed transfer whose receiver is a FHIR server can be repaired as one ward-location update to an existing Encounter, sent with If-Match after a person's approval, and verified from the Encounter's own history. Its own gate needs the server's measured capability, not a declared one.
• S35 plays it on the online demo in Site policy: the case waits for a person's approval before the one update.
• Budget: a cap or share below one run's estimate admits a run while nothing is spent; the install says plainly that OpenAI runs are unpriced, so the caps don't limit them.
• Two unit tests no longer race the running production.

Tested: the full CI on GitHub against IRIS, including the FHIR end to end check, plus the web and browser suites, and a fresh install on the AI Hub build.

Full change log: https://github.com/MxSalata/shutter/blob/main/CHANGELOG.md

Shutter 1.4.6: each case says what the agents contributed, the online demo shows its pathology work, the first screen leads to a short route, and the site profile can ask Shutter about its key answers.

• What the agents contributed, on the case: who proposed the plan (the coordinator agent, one agent, the rules or the site's policy), the plan's recorded reason, what it cites as links to the evidence, and how the specialists stood to it. No agent is named for a plan no agent made.
• Pathology on the online demo: P1 sends a result the results system refuses, and its case goes to a person with no resend; P2 takes the results system down, and one approval has the engine send the three failed results again, unchanged.
• The short route: the trial's welcome and Home both start S2 in Board mode with Run ready; the tour follows S2, S3 and S8.
• Ask Shutter about this, on the site profile: beside a receiver's kind and tier and beside enabling recovery, the Explain agent on IRIS AI Hub explains that answer; asking changes nothing.
• The review on a phone shows what releasing a plan is expected to leave before the confirm button comes into view.
• Hide resolved on Cases; every evaluation figure names its cohort and release; budget admission counts the next run's estimate in one locked step; OpenAI can be configured as a second provider in the installer (not yet measured).
• Docs: two guides (How Shutter works, How to integrate Shutter at your trust), a worked example of a proposed local stop rule, and a two-minute video.

Tested: the full CI on GitHub against IRIS, plus the web and browser suites.

Full change log: https://github.com/MxSalata/shutter/blob/main/CHANGELOG.md

Interoperability

• Stop, start or restart one item of the running production from its row on Production items, as the classic per-item Actions tab does. A stop ends the item's jobs until you start it again: the item stays enabled, messages sent to it wait in its queue, and the next production restart ends the stop. Each one updates the production to apply it, after a warning that says so.
• Copy an item from its row: you name the copy, and it is added disabled, with the source's class and settings.
• Change an item's class, after you type the item's name to confirm. Settings the new class does not have are dropped.
• Business Partners can now be added, edited and removed. A partner that an item still names cannot be removed.
• The agent can propose any of these for you to confirm. Changing an item's class and removing a business partner start switched off; the others start switched on. Change any of them under Agent co-pilot → Governance policy.

Analytics

• A Cube manager screen lists each cube with its facts, its last build and last update, whether it can be synchronized, and the cubes that depend on it. From it you can build or synchronize a cube, and each one is recorded in OcuPilot's audit.
• A build empties the cube, and every cube that depends on it, until the build finishes. A cube that is already updating is refused by name, and so is a synchronize of a cube that was never built.
• The agent can propose a build or a synchronize for you to confirm. Its build action starts switched off and synchronize starts switched on.

Agent

• Investigate a log entry, an alerts.log row included, from the log hub, the error log, the audit log or the log viewer: choose "Investigate this entry", and the agent works through a guided investigation from that entry, returning a recap and its likely causes, ranked, with the rows it used. Like Explain this entry, it changes nothing.

Fixes

• Clicking a name in a list always does something useful. On 17 lists, a name used to lead back to the same list; it now opens that entry's own page, or shows as plain text where there is nothing to open.
• Compiling, deleting and importing in System Explorer, by you or by the agent, now need write access to that namespace's code, as saving and creating already did.
• On Journal records, Newest first now lists the newest records first, and the order you chose is kept when you come back to the page.
• A save sent with a malformed or oversized request now gets a clear error instead of a server error.

Requires InterSystems IRIS or IRIS for Health 2026.2 or later. Upgrading from 1.0.14 keeps your agent definitions, conversations, settings, proposals and audit records.

The demo hours are one window: online without a break until 05:00 UTC on 16 Nov, when voting ends. A Measured table: the AI Hub agent and the fixed rule on 12 held-out cases, two runs, every failure kept. The one-command edition replays the agent's answers recorded in the online demo. In the online demo a visitor answers only the approval task of a sample they sent.The demo hours are one window: online without a break until 05:00 UTC on 16 Nov, when voting ends. A Measured table: the AI Hub agent and the fixed rule on 12 held-out cases, two runs, every failure kept. The one-command edition replays the agent's answers recorded in the online demo. In the online demo a visitor answers only the approval task of a sample they sent.

Shutter 1.4.5 leads each case with what it proposes and opens its review in one press, lets the Cases list hide what is closed, and says how each scenario run executes.

• A waiting case's action button opens the review of the exact plan from any tab, named for what it does ("Approve the resend"). The review leads with the plan's action in plain words and keeps the plan digest, the case version and Cancel in view at the confirming press.
• The Cases list has a Hide closed switch, filtered on the server (GET /cases?open=1), so the total and the pages count the same rows. An empty list fits the install: the trial and the demo offer a scenario, a site says what it watches.
• The Source message link opens the message in the Management Portal: every link names the message header's class.
• Scenarios shows each run mode's words in its own card, says how a run executes apart from who proposes, and leads a finished run to its own case. Home starts a short route: S2, then S3, then S8.
• Provenance in plain words: the trial's runs say "Browser recording: no new inference", and a run is called live only when its own case shows a model call. S8's "every check passed" says that Rules only is exempted there on purpose and that the gate holds the resend on a real case.
• The site profile's questions line up, and the Orders and results card says what a case gets and how the outage resend works.
• The repository checks run in their own workflow on every push to main, documentation-only pushes included.
• The README puts S8 side by side under the opening, has one block for judges, gives each figure its cases, runs and release, and ends with a note on how Shutter was developed. The install guide makes enabling recovery its own step after the conformance and stop tests.

Tested: the full CI on GitHub against IRIS, plus the web and browser suites.

Full change log: https://github.com/MxSalata/shutter/blob/main/CHANGELOG.md

The agent and the fixed rules, measured side by side

• The README now sets OcuTalon's agent beside its own fixed rules, the fallback it uses when it has no model key, on the same recorded evaluation cases: 24 routine items and 11 incidents, plus 12 held-out cases (8 routine items and 4 incidents) whose expected answers were committed before either side ran on them. See With the agent and with fixed rules only.
• With Claude Haiku 5.5 at this release's evaluation, the agent got 22 of the 24 routine items, 11 of the 11 incidents and 12 of the 12 held-out cases right, and it put one item that needed you now in the digest instead. The fixed rules alone got 6 of 24, 0 of 11 and 2 of 12; they tell you about everything, so they missed none.
• The README's Measured table now comes from this release's evaluation too.

Routine attention decisions

• When an item of a watched production is disabled, OcuTalon now decides Tell now instead of Act. In this release's evaluation it got 22 of 24 routine attention decisions right; the README's earlier figure, from release 1.0.4, was 21 of 24.

Sign in to your mailbox with Google or Microsoft 365, on your own install

• An admin can now connect the mailbox in Settings > Channels > Email by signing in with Google or Microsoft 365, instead of saving a password. OcuTalon then reads the mailbox over POP3 and sends over SMTP with the sign-in, and IRIS renews the sign-in's access token when it expires.
• If the sign-in is withdrawn at the provider, the next check of the mailbox says "The mail sign-in is no longer valid." and Settings offers the sign-in again.
• The guide has the steps for both providers: Gmail and Microsoft 365. The online demo has no real mailbox and no sign-in.

Your daily digest by email, on your own install

• With an outgoing mail server saved, Settings > Daily digest can also send your digest to the connected mailbox's own address. OcuTalon doesn't read its own digest back as a new email.
• Mail is off on the online demo.

A rejected request leaves the Status section at once

• When you reject an approval request, it now leaves the Approval requests list in the Status section at once, instead of a second or two later. Try it on the online demo: after the guide's incident, More to try, Reject a fix.

The README names the platform parts

• The judges' table now names the AI Hub and Interoperability parts OcuTalon is built from: its three %AI.Agent agents, the outbound adapter, the AI Hub classes behind the tool gate and Visual Trace, the RAG knowledge base, the Config Store entries, the EnsLib.Workflow approvals and the resend that a fix uses.

Requires the InterSystems IRIS for Health AI Hub Early Access image, build 154 or build 109. Upgrading from 1.0.8 keeps your conversations, reminders, watches, approval requests, settings, memories, documents and Telegram pairing. The online demo runs 1.0.9.

---

Online demo: https://demo.ocutalon.com · Guide: https://github.com/jbrandtmse/OcuTalon/blob/main/GUIDE.md · Open Exchange: https://openexchange.intersystems.com/package/OcuTalon

The tagline names the IRIS entry. The judge page adds tonight's demo hours, says what the page shows when the model service does not answer, and says that build 109 has not been tested.

The listing is now the IRIS judge page (iris/README.md): the online demo with its hours, the 43 s live-run video, Judge it in 2 minutes and the contest task part by part. The online demo shows no error line, and each decision's trace is on the page with no login.

Vigil 1.0.3

Fixes from the Open Exchange quality review:

• Fixed all DynamicObject fields with underscores: every class now uses %Get and %Set (for example scoreObj.%Get("final_level")), including Vigil.BP.VigilProcess. All classes compile cleanly.
• The app now starts out of the box: docker compose up -d uses the public IRIS for Health Community image (intersystemsdc/irishealth-community). No private image, no tarball and no API key needed.
• Fixed a first-boot stop on the Community image, so IRIS stays running after the first start.
• The ward board opens at http://localhost:52773/vigil/ and also at http://localhost:8000/ (port 8000 now works).

Feature improvements:

• The consent queue now shows the seeded consent reviews.
• Night Shift replays all 118 seeded events (74 ADT, 40 lab results, 4 referrals) and shows the count on the board.
• Live mode (optional, needs your own Gemini key) now falls back safely to the clinical safety floor if the model is unreachable or the key is invalid. No errors show on the board.

Docs:

• The README quick start is now: git clone, docker compose up -d, open http://localhost:52773/vigil/
• The README now says clearly which demos are replay or scripted, and that live AI is optional.

Tested: fresh clone with docker compose up -d, health check healthy (version 1.0.3, replay mode), ledger verify and tamper test passing, 14 of 14 automated tests passing.

KnAIghts of The Round v0.2.0 expands the source release with new tools for governed interoperability workflows.

• Inspect production message classes and declared routes.
• Review synthetic HL7 edits alongside their transport receipts and proposed FHIR observations.
• Generate OpenAPI descriptions for selected FHIR R4, R4B, and R5 read operations.
• Connect Workflow review, a custom adapter, Java PEX, and IRIS Vector Search in the supported synthetic profiles.
• Explore the live demo, where IRIS validates agent proposals and records the resulting disposition and receipt.

The public IPM package remains the qualified 0.1.0 core. The expanded integrated package has been tested locally and is being prepared separately for public distribution.

Update test script

Update deployment version

Initial Release

DICOM3D_AI v0.2.0

DICOM3D_AI is a research prototype connecting source-linked imaging evidence to an InterSystems IRIS AI Hub interoperability workflow. A native agent reviews permitted evidence, a deterministic validator checks the structure identity and exact citations, and the viewer can show the corresponding finding. Conflicting synthetic evidence follows a review-request path.

What’s new

• FHIR to OpenAPI: An offline tool projects supported read and search declarations from FHIR R4, R4B and R5 CapabilityStatements into OpenAPI.
• Source-linked evidence: Experimental checks compare H000 study, imaging-instance and segmentation identities before accepting a finding. R4 and R4B projections remain research-only and review-only.
• Interoperability metadata: A separate synthetic production demonstrates passing and checking request, case, study and evidence identifiers.
• Test assistance: Optional local tools support reviewed declarative plans for fixed synthetic test cases.
• Safe Demo accessibility: Focused keyboard and focus-handling improvements make selected panels easier to navigate.
• IPM packaging: The module includes the native workflow classes and selected research tools. The viewer, FHIR service, model, provider, credentials and production require separate setup.

Scope and limitations

The browser-only Safe Demo uses simulated AI outcomes; it does not run the native agent. Historical native executions and a separate experimental R5 viewer-selection witness are documented in the repository. Intermittent inference failures and incomplete repeatability remain known limitations.

This release is for research and demonstration. It is not clinically validated, and the generalized anatomy is not registered to patient imaging. FHIR projections do not establish complete FHIR conformance or access to a live server.

See the installation instructions and judge walkthrough for prerequisites, evidence and setup boundaries.

A shorter README, and a guide for everything else

• The README is now a short page for judges and first-time visitors: the online demo, where to look for each judging criterion, one incident end to end, what's interesting, what was measured, and the contest checklist, with the full video (3:04) linked at the top.
• Install, features and administration moved, word for word, into a new OcuTalon guide. The page's install, model-key and Telegram links, and the install script's failure help, now point to it.

Webhooks, on your own install

• An admin creates a webhook in Settings, under Channels, with a name and a secret. Any other system can then send it a JSON event with POST, with the secret in a header.
• OcuTalon answers at once, then decides on the event as untrusted content: it can lead to a notice or to your digest, and nothing in the event can approve, change or run anything. The decision shows on the History page, and the event and its decision in Visual Trace. The secret is kept only as a salted hash.
• Webhooks are off on the online demo, which has no public webhook address. Try it on your own install: More to try, Send an event through a webhook.

Your own mailbox, on your own install

• An admin connects a POP3 mailbox in Settings > Channels > Email. OcuTalon makes an attention decision on each email that arrives after that, sends the email to the model for it, and leaves your mail on the server.
• Email is read as untrusted content: nothing in an email can approve, change or run anything. The online demo keeps its simulated mailbox.

Queue backlogs between digests

• Ask OcuTalon in the chat to watch the queue backlogs, and every minute it reads the queues of each production it can read. When a queue holds more than 100 messages, or its oldest message has waited over 10 minutes (an admin can change both), OcuTalon decides once whether to tell you now, and again only after the queue has dropped below and backed up again.
• The check only reads, and a run that can't read a namespace shows Not checked, never Passed. Try it: More to try, Watch queue backlogs.

Partner notices say only how the incident ended

• When you ask OcuTalon to warn the partner, its own code now adds how the incident ended to the draft's facts, from the incident's records: for example "all 50 affected orders went through after the fix", or, after a reroute, that they "were resent by another route". When the records don't match one of those endings exactly, the notice doesn't say whether the orders went through.
• A notice for a resolved incident no longer says that work is still going on. On the online demo the simulated partner contact is now Orders Partner IT. Try it: after the guide's incident, More to try, Warn the partner.

A suspended item keeps its incident open

• On your own install, an incident whose fix suspended an item now stays open until you ask OcuTalon to resume the item, or dismiss the incident yourself, so you can still undo the suspend the governed way: an approval request with its own dry run.
• If you dismiss it while the item is still disabled, the next check reports the disabled item once, with the time of the suspend. Suspend and resume stay refused on the online demo, where every visitor shares the Orders production.

More to try

• The guide's More to try panel now has 16 items, 15 of them on the online demo.

For developers

• The evaluation can now measure the fixed rules alone, with no model and no key: eval/run.sh --rules-only, on the main set of cases and on a held-out set.

Requires the InterSystems IRIS for Health AI Hub Early Access image, build 154 or build 109. Upgrading from 1.0.7 keeps your conversations, reminders, watches, approval requests, settings, memories, documents and Telegram pairing. The online demo runs 1.0.8.

---

Online demo: https://demo.ocutalon.com · Guide: https://github.com/jbrandtmse/OcuTalon/blob/main/GUIDE.md · Open Exchange: https://openexchange.intersystems.com/package/OcuTalon

1.0.1

Added

• Online demo support (InterSystems DevCloud): the Dockerfile now defaults to the AI Hub build 154 image that DevCloud can pull, and sets the non-secret demo settings (SN_SIM_BASE_URL, SN_BACKGROUND_TASKS=1), so it runs with no .env. Local installs are unchanged: docker compose still builds on the image you load and retag.
• "Signal Noir at a glance" in the README: an infographic and a short text walkthrough of the six steps, from a failed Order to the partner's confirmation.
• Logo in the operator console's top bar and as its favicon.
• Developer Community article (English version), with code, console screens and the infographic.

Changed

• The README and the article link the repository by its new name, musketeers-br/signal-noir.

Notes

• No change to the Agent, the Validator or the API: the frozen reference run (agent-v5, gpt-4o-mini, rules-v2) still applies.
• The provider key is never set in the image. Without one, Cases go to review with a Dossier that says the provider is unavailable.

• Replaced the Python PEX name comparison operation with a Java PEX implementation.
• Added native Interoperability message tracing for AI model and tool calls.
• Updated package resources and installation setup for the Java operation and tracing components.

• Replaced the Python PEX query embedding operation with a Java PEX implementation using the pinned ONNX model.
• Added native Interoperability message tracing for AI model and tool calls.
• Updated package resources and installation setup for the Java operation and tracing components.

add Deployment version

Listing update: the video field now links the YouTube demo and the MP4 (the old field joined two URLs). The package code is unchanged from 0.1.9 on the community registry.

Shutter 1.4.4 lets anyone on the online demo play the scenarios, from their recordings.

• The public demo's account can press Run on Scenarios: every agent of its cases replays its recording, so no model is called and nothing counts against the demo's budget. The judges' runs stay live.
• A replayed run says so on its case, its agent runs and its banner. The single investigator, never recorded, is greyed out for it.
• Reset stays locked for the public account; the hourly reset cleans up after visitors.
• The docs name the two recovery paths and what each guarantees: a write through the receiver's API is verified at the receiver; an outage's resend through the engine is settled by the receiver's declared duplicate handling.
• A demo video: 13 minutes recorded live on the online demo, with captions, on YouTube and attached to the release as an MP4 that downloads without signing in.
• The README leads judges through the whole workflow, with links to the classes that carry it, and puts the measured results first: all twelve held-out cases rerun on 1.4.3, with no wrong write executed in any mode.
• The README says the judges share one account on the demo, and with it 2 USD of live runs an hour; when that is spent, the demo account still plays the scenarios from their recordings.

Documentation corrected since the release (no code change):

• S8's comparison names the one gate check it lifts, ENTRY_NOT_NAMED_WRITE, for a recorded scenario in Rules only; a real case gets the whole gate.
• How a case is released, in order: the receiver's dry run comes before the approval, which binds that dry run, the plan, the evidence, the policy and site profile versions and the receiver's version and epoch, and expires after an hour.
• A hospital's install commands switch the demo off (-Ddemo=0), and a new site install starts in shadow mode, which sends no recovery write and no escalation email.
• Turning off a receiver's answer checks (strict_answers: false) is described as a weaker contract, not a fix.
• Documents, outages and receivers: the switches a document exception needs, Outage Mailer and Read again, and one mapped operation through the receiver's API.
• The results state what each figure measures, release by release.

Tested: the full CI on GitHub against IRIS, plus the web and browser suites.

Full change log: https://github.com/MxSalata/shutter/blob/main/CHANGELOG.md

Documentation, recorded evidence and judge-tooling update. No native code changes. The keyless launcher now pins verified native 0.3.2 explicitly, so publishing a newer package cannot move its install target. The local candidate verifier supports an offline literal install and checks installed identity. The recorded held-out wrong-delivery/review trade-off and three source-linked prose examples are easier to inspect. Includes the recorded evidence tour and receipt inspector. Native package 0.3.3; separate React workbench 0.2.1. Agent, deterministic policy, runtime sources and frozen evaluation records are unchanged; no new evaluation was run.

ISAIAH 0.2.0 — Developer Tools and FHIR Interoperability

ISAIAH began with a principle: an agent's answer isn't enough. We need to understand what it observed, verify its conclusions, and retain evidence of what actually happened.

Version 0.2.0 extends that approach beyond the original R17/R29 interoperability investigations into developer tools, interface inspection, and a controlled HL7-to-FHIR workflow.

What's New

• FHIR R4, R4B, and R5 OpenAPI export. Generate standalone OpenAPI 3.1 documents from approved local FHIR CapabilityStatements. The exporter includes advertised resource read/search operations, identifies omitted capabilities, and keeps advertised behavior separate from verified endpoint behavior. This implements the read/search portion of Community Idea DPI-I-804.
• Version-aware HL7-to-FHIR delivery. The generated synthetic Organization application was qualified against isolated local HAPI endpoints for R4, R4B, and R5. It binds delivery evidence to the destination, rejects version mismatches and conflicting resources, and handles duplicate requests and uncertain responses without blindly repeating writes.
• AI-assisted test proposals. Native AI Hub can propose bounded synthetic test inputs. ISAIAH derives expected results from a reviewed contract; model output does not directly become executable code or deployment authority.
• Interface and message-type inspection. Compare declared ObjectScript message types with independently observed trace information and generate reviewable, versioned synthetic interface artifacts.
• Evidence-linked developer guidance. Coach, Focus, and Companion present the same underlying facts while adapting the level of explanation.

Installation and Qualification

The 0.2.0 native IPM package passed clean installation on IRIS for Health Community AI Hub 2026.3.0AI Build 154U. A targeted upgrade from public IPM 0.1.0 to the qualified 0.2.0 archive also preserved existing synthetic receipts and message history.

The separate Node workflow passed clean-install testing and local R4/R4B/R5 delivery checks. The retained release review reports 44 passing offline tests, with two optional Hermes checks skipped.

Current Boundaries

ISAIAH remains an open-source local developer preview using synthetic data, not a production or clinical integration system.

The generated Organization workflow is intentionally narrow. OpenAPI export describes advertised capabilities; it does not establish access permissions or prove live endpoint behavior. Shared or untrusted writable project directories, general Node/Compose migrations, and unrestricted application generation are not qualified.

The public interactive demonstration is a simulation. The downloadable R17/R29 recording shows genuine earlier native execution, with its recording limitations disclosed.

Explore ISAIAH

• GitHub repository and installation instructions
• FHIR OpenAPI walkthrough
• Developer workflow documentation
• Interactive simulation
• Media gallery

Developed under the direction of Kenneth J. Smith with AI-assisted development and evidence-based verification.

SynapseHealer v1.0.1

🚀 Highlights

1. Native InterSystems Workflow Engine Routing

• Role Delivery: Created and verified the SynapseReviewers workflow role via Synapse.Config.WorkflowSetup, populated with authorized clinician/operator users.
• Two-Way Action Handling: Native Approve and Reject actions executed in the IRIS Workflow Inbox now invoke Synapse.Workflow.ReviewTaskResponse to transition local queue states and dispatch payloads to ResubmitDeliveryOp or QuarantineVaultOp.
• Portal Integration: Added deep-link navigation directly from the dashboard into the native InterSystems Zen Workflow Worklist (/csp/sys/%25CSP.UI.Portal.WorkflowWorklist.zen).

2. IPM (InterSystems Package Manager) Certification

• Verified module.xml specification for automated registration on the IPM Community Registry.
• Removed packaging artifacts that caused zpm load failures, enabling standalone installation via:

  zpm "install synapse-healer"
  

3. 24-Point Automated Verification Suite

• Added scripts/smoke_test.sh providing 24 automated checks across:
  • Idempotent workflow setup.
  • Role existence and active reviewer memberships.
  • Interoperability production runtime checks (Synapse.Production).
  • Task creation, task claim, approval, and rejection.
  • End-to-end execution of all 7 canonical HL7 and JSON replay scenarios.

4. Observability Dashboard Polish (ui/)

• Pipeline Stepper Animation: Visual step-by-step illumination across the 4 stages ($1 \rightarrow 2 \rightarrow 3 \rightarrow 4$) with smooth auto-scroll to the Diff Studio.
• Diff Token Badging: Exact mutated tokens glow in rose (1985-02-31, unescaped |, quoted strings) vs emerald (19850228, \F\, native types).
• HL7 Segments: Segments (MSH, EVN, PID, PV1, OBX) are split onto individual numbered lines.
• Presenter Reset: Added a one-click Reset control in the top bar to zero out counters on camera for video demos.
• Realistic Telemetry: Realistic latency variance derived from individual tool runtimes replacing flat static metrics.

---

Full Commit Diff: v1.0.0...v1.0.1
License: MIT
Author: Joseph Bassey John [email protected]

Suspend, resume and reroute, with a dry run

• Ask OcuTalon in the chat to suspend or resume an item of a watched production, or to reroute an incident's failed messages to another item. Each comes as an approval request with its own dry run: the item's state before and after, or the count, the current and new target and one example.
• Nothing changes until you approve, and after you approve the request's card shows what happened.
• While the incident is open, a suspend is undone the same way, by asking OcuTalon to resume the item. After it closes, enable the item again on the production's configuration page in the Management Portal.
• On the online demo every visitor shares the Orders production, so there OcuTalon offers the reroute only. Try it: after the guide's incident, More to try, Reject a fix, then Reroute the failed orders.

Partner delay notices, sent only after approval

• Ask OcuTalon in the chat to warn the partner about an incident's delay. It drafts a short notice from the incident's count, times, affected flow and cause, never from a message's contents, and raises it as an approval request that shows the exact text and every recipient.
• Nothing is sent until you approve. In demo mode the partner contacts are simulated and the approved notice goes to the Simulated mailbox, not to a real inbox.
• Try it: after the guide's incident, More to try, Warn the partner.

The Memory card: pinned and learned memories

• All memories lists what OcuTalon remembers, each marked Told or Learned, with Pin, Unpin and Forget on each and a search box.
• Pinned shows the facts read at the start of each conversation and how much of their 1,000-token limit (an estimate) they use.
• When OcuTalon keeps a fact you didn't ask it to remember, OcuTalon's code adds a line after the reply in the web chat and in Telegram: I'll remember this: "…". You can tell me to forget it. The line doesn't depend on the model.
• Try it: More to try, Pin a memory.

A Usage page

• Open Usage in the header to see your tokens and cost by day, agent, channel or model, counted per turn and per decision, with each day's largest model request (an estimate) against the 100,000-token limit.
• On the online demo you see only your own session's use. Try it: More to try, See tokens and cost.

A History page

• Open History in the header to look back at the decisions OcuTalon made on your messages, filtered by outcome, channel, date and fallback, each with its reasons and a link to its trace, and at your incidents with their state and, once closed, their summary.
• On the online demo you see only your own session's decisions and incidents. Try it: More to try, Look back at past decisions.

More to try, in four groups

• The guide's More to try panel now has 14 items under four headings: Incidents and fixes, Memory and documents, Safety and settings, and Usage and history.

Docs and CI

• The README's logo now uses a full web address, so it shows wherever the README is read.
• CI's browser and install tests now wait for a busy test port and retry once it is free, instead of failing.

Requires the InterSystems IRIS for Health AI Hub Early Access image, build 154 or build 109. Upgrading from 1.0.6 keeps your conversations, reminders, watches, approval requests, settings, memories, documents and Telegram pairing. The online demo runs 1.0.7.

---

Online demo: https://demo.ocutalon.com · Open Exchange: https://openexchange.intersystems.com/package/OcuTalon