ConnectHealth v0.3.74
The CharmHealth integration, workflows that query your own data lake and wait on parallel work, and mappings that move between environments.
This release adds CharmHealth to the supported EHRs, lets a workflow query data already sitting in your own AWS storage and wait for several parallel branches before continuing, and makes mappings portable between environments — a workflow you export now carries its mappings with it. A new Mapping Academy documents the mapping engine end to end, and API keys are now masked by default.
✨ What's New
🏥 CharmHealth Integration ⭐ Highlight
CharmHealth joins the supported EHRs as a first-class connection, set up the same way as every other integration in the Workflow Builder.
- ✅ Every launch style — connect on its own, embed inside the EHR, or run unattended system-to-system
- ✅ CharmHealth SSO — its single sign-on launch is supported and verifiable from the connection test page
- ✅ Nothing new to learn — the same connection screens, the same workflow steps as any other EHR
🔎 Query Your Own Data Lake From a Workflow
A workflow can now run a query against data you already keep in AWS storage and use the result in the next step — reporting, reconciliation or enrichment without moving the data anywhere first.
- ✅ Parameterized queries — the query takes values from the message flowing through the workflow
- ✅ AWS credentials declared once — one shared AWS setup serves every AWS-backed step in your flows instead of repeating credentials on each one
- ✅ One place to find them — all AWS steps now sit together in a single group in the builder palette
⏸️ Wait for Parallel Work to Finish
A workflow that fans out into several branches can now wait for all of them and continue with one combined message — useful when a decision needs answers from more than one system.
- ✅ Collect and continue — wait on several concurrent calls, then proceed once
- ✅ Choose how results combine, or take the sensible default
- ✅ Wiring mistakes are reported up front rather than showing up as a workflow that quietly hangs
📦 Move Mappings and Workflows Between Environments
Export a mapping project to a single file and import it elsewhere — test into production, or between accounts.
- ✅ A workflow export brings its mappings — every mapping the workflow uses travels with it, and an import restores them together
- ✅ All or nothing — name clashes are detected and resolved before anything is written, so an import never leaves a half-loaded project behind
- ✅ You decide on conflicts — keep what is there, or bring the incoming copy in under a distinct name
📚 Mapping Academy
A complete usage manual for the mapping engine introduced in v0.3.73 — nine guides plus three working examples you can download and open.
- ✅ Guides — Quick Start, any-format-to-any-format reference, Auto-Map, conditions, repeating groups, filters and templating, export and import, and using a mapping inside a workflow
- ✅ Working examples — a starter mapping, FHIR Observation to HL7 results, and an X12 eligibility response to FHIR coverage
- ✅ Short screen recordings for the quick start and auto-map walkthroughs
🔐 API Keys Masked by Default
Listing your API keys no longer shows their values. Reveal a single key when you actually need it.
- ✅ One key at a time — a list view can no longer expose every secret in the account at once
- ✅ Sessions renew themselves — an embedded session can be renewed by presenting itself, with no API key involved, so a clinician is not interrupted mid-encounter. A session renewal credential from a different launch is refused
🔧 Improvements
- Helix chat history survives restarts — conversation history now lives in durable storage rather than on the running container, so it persists across restarts and stays consistent however many instances are running
- Your search terms stay private — a term typed into Helix search is removed before any error is recorded
- Workflow run traces get their own encrypted log stream, retained for 90 days by default and adjustable to match your operational policy
🐛 Fixes
| Area | Fix |
|---|---|
| HL7 in a workflow | A message arriving from an HL7 connection now feeds the parser and mapping steps directly — no placeholder text, and no intermediate step needed to unwrap it |
| Workflow web endpoints | A valid session is now accepted on a workflow's own HTTP endpoints; a refused request records why |
| Mapping — renamed fields | Renaming a target field keeps its mapping instead of dropping it, with no leftover duplicate row when a mapped field becomes a list |
| Mapping — naming | A duplicate project or template name is flagged inline as you type |
| Connections | Required setup details on an Open Dental connection are now marked as required |
| Helix | A failed request reports its actual cause instead of a generic error |
📝 Upgrading
Deploy this release using the infrastructure templates included with it. Two things to know before you do:
- You now choose the initial administrator password at deploy time. It is required, there is no
shared default any more, and the deployed stack no longer publishes a password value of its own.
- Anything that read API key values from the list response must switch to revealing the single
key it needs.
Existing Helix conversation history on a running container is not carried over; new conversations are stored durably from the first start on this version. No database migration is required, and existing connections, workflows and mappings keep working as they do today.
🚀 ConnectHealth v0.3.74 — CharmHealth, data-lake queries and parallel waits in workflows, portable mappings, and the Mapping Academy