Early Access · beta. Help test and shape the 1.0 release. Where the project stands →

MessageFoundry™
The toolkit

Tools and commands

The toolkit includes the engine, console, editor, command-line tools, and migration relay. Use the list below to find each tool and its command.

One contract, many clients

Tools connect through the engine API

The engine owns the message store. Operator tools connect through its local API; the migration tee relay sits in front of the engine.

The toolkit — one engine, many clients A central Engine service exposes one localhost API. Four tools — the command-line tool, the admin console, the VS Code extension, and the test harness — are separate processes that talk to it over that API. A tee relay sits in front of the engine, forwarding bytes during migration. in front localhost API the one contract · HTTP / WS Engine service headless runtime owns the store & staged queue Command-line tool messagefoundry <cmd> serve · check · generate Admin console dashboard · replay delivery & audit trail VS Code extension wizards · test bench Stage → Promote · AI Test harness python -m harness synthetic, PHI-free load Tee relay sits in front of a legacy + a shadow engine, forwards the same bytes to both · test only
Each tool, up close

How you run it, and what it's for

Engine service

messagefoundry serve

Runs Connections, Routers, and Handlers through the staged queue and owns the store. Exposes a local HTTP/WebSocket API. Supports Windows service installation through NSSM.

Command-line tool

messagefoundry <cmd>

Use serve to run the engine and init to create a config repository. validate, dryrun, and check form the commit and continuous integration gate. Use connection to edit connections.toml and generate for synthetic HL7. The tool also handles keys and audits. Git hooks and the VS Code extension run the introspection commands, which make no network calls.

Admin & monitoring console

open http://127.0.0.1:8765/ui

Use the browser console to view connections, search messages, and inspect delivery and audit records. It includes replay, an HL7 parse tree, dead letters, alert rules, and user and session controls. It connects through the API and never accesses the database directly. Tour the console →

VS Code extension

open in VS Code → F5

Write routes with HL7 autocomplete, validation on save, and the New Route Wizard. The Test Bench runs .hl7 files with before-and-after diffs and step-through debugging. Use Stage → Promote to deploy to a running engine, or optional @messagefoundry AI chat for coding help. Tour the editor →

Test harness

python -m harness

Test a running engine with synthetic traffic containing no patient data. Send, Receive, File, Compose, and Monitor tabs let you inject acknowledgment faults, malformed messages, and delivery failures. Headless scenarios check message outcomes in continuous integration. A separate load-testing engine supports warmup, ramp, and soak profiles, with a service-level objective report. Install it with pip install messagefoundry-harness. It ships separately from the engine wheel, with matching releases.

Tee relay

python -m tee

Place the tee relay before a legacy engine and a shadow MessageFoundry instance. It acknowledges receipt and forwards the same bytes to both engines for comparison. Stop the relay to roll back. Use test and synthetic data only; the relay is not hardened for protected health information. Run it from an engine repository checkout; it is not published as a package.

The engine is the only long-running service and the only process that accesses the store. The console, extension, and harness use its localhost API; the tee relay sits in front. That API handles authentication, role-based access control, and audits. See the architecture →

Get started See how it fits together →