medicafe 0.260701.3

MediCafe

MediCafe

MediCafe is a mixed-runtime healthcare workflow automation repository. It combines the current XP-compatible clinic operator system (MediCafe, MediBot, MediLink, and xp_client), a local shadow/readiness path that validates future architecture against local evidence, and a Python 3.11+ GCP orchestration path in cloud/orchestrator.

This repository intentionally supports more than one runtime model. It is not a single deployable app with one entrypoint.

Runtime And Evidence Layers

Layer	Primary code	Python	Primary use	Packaging posture
Production XP/local operator system	MediCafe, MediBot, MediLink, xp_client	3.4.4-compatible package path	Clinic-side claims, eligibility, deductible, reconciliation, download, Medisoft-oriented automation, and XP-side cloud queue consumption	Packaged by setup.py for PyPI / XP install
Local shadow validation system	scripts/unified_model, selected MediCafe/cloud_readiness* helpers	3.4.4-compatible local/package path	Shadow A→F pipeline, local relational validation, lineage, parity, Phase C/D/F evidence, and future-process proof	Selected scripts are packaged because XP invokes them by file path
Local cloud-readiness / evidence layer	MediCafe/cloud_readiness*, MediCafe/launcher_cloud_readiness_mixin.py, support/evidence helpers	3.4.4-compatible local/package path	Operator system-health packs, run-evidence bundles, Layer 1 Readiness, recommendations, post-update validation, and support-send workflows	Packaged when used by XP/local launcher and support flows
GCP cloud orchestrator	cloud/orchestrator, cloud/orchestrator/tests	3.11+ repo-run / GCP path	Gmail ingestion, queueing, preprocessor validation, alerting, setup, runtime operations, and future cloud deployment work	Source lives in git for dev/GCP work; it is not part of the XP/PyPI package payload

setup.py packages the XP/local operator system plus XP-required shadow/readiness files. The GCP cloud orchestrator is maintained as repo-run Python 3.11+ source and should remain inert with respect to the clinic-side PyPI release.

Primary Entry Points

Package CLI:

medicafe launcher
python -m MediCafe launcher

Current routine package commands exposed by MediCafe/__main__.py:

• medicafe launcher
• medicafe medibot [config_file]
• medicafe medilink
• medicafe claims_status
• medicafe deductible
• medicafe download_emails
• medicafe cloud_daemon [config_file]
• medicafe send_error_report
• medicafe send_queued_error_reports
• python -m MediCafe cleanup_local_artifacts
• medicafe docx_index_rebuild
• medicafe preflight
• medicafe reconcile
• medicafe version

Support, analytics, and local evidence commands exposed by MediCafe/__main__.py:

• medicafe send_cpt_dx_review
• medicafe send_minutes_charge_review
• medicafe build_service_line_charge_anchor_index
• medicafe local_failure_pattern_analyzer
• medicafe layer1_local_diagnostics
• medicafe docx_index_retry_failed
• medicafe csv_metadata_index_rebuild
• medicafe analyze_csv_excluded_insurance

medicafe deductible is kept as the advanced/compatibility checker. The normal operator path for the consolidated deductible form is inside MediLink Claims as Open deductible report (United, 90-day DOS).

Canonical cloud operator path:

py -3.11 cloud/orchestrator/validate_and_complete_setup.py

Repository Map

• MediCafe/: shared package, CLI entrypoint, launcher, preflight, cloud-readiness helpers (including Layer 1 Readiness for operator system-health and run-evidence bundles; see MediCafe/layer1_readiness_model.py and AGENTS.md), API utilities, reconciliation, error reporting, and shared runtime helpers
• MediBot/: Medisoft-oriented local preprocessing and data-entry support scripts
• MediLink/: claims, eligibility, deductible, remittance, Gmail helper, and Apps Script-related assets
• cloud/orchestrator/: Python 3.11+ Cloud Run service, auth/config/runtime helpers, preprocessor gate, watch/admin flows, setup scripts, and cloud tests; repo-run/dev/GCP source, not XP package payload
• xp_client/: XP-side cloud queue consumer (medilink_cloud_daemon.py)
• tests/: main regression suite for package/local workflows, API helpers, launcher behavior, preflight, cloud-readiness, and integration coverage
• scripts/unified_model/: XP-compatible local shadow-pipeline and unified relational validation scripts; selected scripts are packaged as runtime data files because XP flows invoke them by path
• sql/: SQL artifacts used by migration and validation work
• docs/: maintained documentation surface
• tools/: operator, build, audit, and verification helpers; only XP-required tools should be treated as runtime-critical

The workspace root also contains operational inputs, artifacts, and generated state such as input/, TEST_CLAIMS/, ERA_TEST_DUMP/, reports_queue/, archive/, build/, dist/, .pytest_cache/, and tmp*. Do not treat those folders as the canonical source layout.

Local artifact cleanup:

python -m MediCafe cleanup_local_artifacts
python -m MediCafe cleanup_local_artifacts --apply --purge-queue

The cleanup command preserves README markers, defaults to dry-run, and skips queued ZIP bundles unless queue purge or queue-age retention is explicitly requested.

Automatic cleanup runs only from a development checkout (.git present) and only when MEDICAFE_LOCAL_ARTIFACT_AUTOCLEAN is not disabled. It is checked at python -m MediCafe ... startup and before support bundle creation. Defaults: at most once every 6 hours; apply cleanup when generated artifacts exceed 200 files, 250 MB, or 50 queued ZIPs. Ordinary generated artifacts and queued ZIP bundles are pruned immediately once a threshold is crossed. Set MEDICAFE_LOCAL_ARTIFACT_AUTOCLEAN_QUEUE_MAX_AGE_DAYS to a positive value only if a machine needs delayed queue retention.

Configuration

Primary local configuration files:

• json/config.json
• json/crosswalk.json
• json/medisoftconfig.json

Cloud/orchestrator runtime settings are managed through cloud/orchestrator/config.py, environment variables, and the validator/setup flow.

Testing

Main test surfaces:

• tests/
• cloud/orchestrator/tests/

Canonical automated test entrypoint:

python3 -m pytest -v

Windows equivalent:

py -3.11 -m pytest -v

XP/Python 3.4.4 local test environments should install the runtime pins from requirements.txt when needed, then use the package test extras for the legacy pytest pins before running pytest-driven files:

py -3.4 -m pip install -r requirements.txt
py -3.4 -m pip install ".[test]"
py -3.4 -m pytest -q tests/test_json_io.py

pytest.ini collects tests/ and cloud/orchestrator/tests/, with manual excluded from recursion. For focused runs:

python3 -m pytest -q tests/test_preflight.py
python3 -m pytest -q tests/test_config_loader_diagnostics.py
python3 -m pytest -q tests/test_error_reporter_configloader_degraded.py
python3 -m pytest -q tests/test_launcher_next_step_orchestration.py
python3 -m pytest -q tests/test_cloud_readiness_health.py
python3 -m pytest -q tests/test_cloud_readiness_artifacts.py::TestLayer1ReadinessModel tests/test_cloud_readiness_artifacts.py::TestLayer1ReadinessSurfacesAndContract tests/test_cloud_readiness_artifacts.py::TestSystemHealthPackWiring::test_send_latest_system_health_pack_extra_meta_merges_layer1_readiness_shallow_keys tests/test_cloud_readiness_artifacts.py::TestRunCentricArtifactTools::test_send_latest_run_evidence_bundle_extra_meta_merges_layer1_readiness_shallow_keys tests/test_cloud_readiness_recommendations.py
python3 -m pytest -q tests/cloud_readiness/test_self_test_bundle_build_telemetry.py
python3 -m pytest -q cloud/orchestrator/tests

See docs/TESTING_GUIDE.md for the current suite shape, selector lanes, default excludes, and standalone script-style harnesses that are not part of the canonical pytest run.

Documentation

Start here:

• docs/README.md
• docs/MEDICAFE_MASTER_GUIDE.md
• docs/TESTING_GUIDE.md
• docs/MEDICAFE_API_ARCHITECTURE.md
• docs/MEDILINK_CLOUD_MIGRATION.md
• docs/architecture/CSV_DOCX_DAT_JOIN_SCAFFOLDING_INVENTORY.md (shadow CSV/DOCX/DAT correlation, cloud readiness read path, orchestrator context)
• cloud/orchestrator/README.md

Security and Repo Hygiene

• Avoid PHI in logs, reports, and committed artifacts.
• Use send_error_report and send_queued_error_reports for diagnostics instead of ad hoc data exports.
• Treat tokens, certificates, raw client secrets, build outputs, and downloaded artifacts as operational state, not documentation or source of truth.

License

MIT License. See LICENSE.