interoperability-serving
Router for OpenMed CLI, REST, gRPC, MCP, adapter registry, and interoperability handoffs.
- License 许可
- Apache-2.0
- 最近更新
- 2026-09-03
- Stars
- ★ 351
- Forks
- 27
01 前言 Front Matter
- name
- interoperability-serving
- license
- Apache 2.0
- metadata
- { "disco-role": "operating" }
- description
- Router for OpenMed CLI, REST, gRPC, MCP, adapter registry, and interoperability handoffs.
- disable-model-invocation
- true
02 资源文件 4
- 参考文档references/cli-reference.md4.0 KB
- 参考文档references/interop-serving-workflows.md6.2 KB
- 参考文档references/troubleshooting.md6.1 KB
- 脚本scripts/openmed_cli_probe.py9.2 KB
03 SKILL.md 正文
Interoperability and serving
Use this sub-skill for:
- one-shot CLI validation and transport-safe probes;
- REST, gRPC, or MCP service use;
OpenMedClient,create_app(), andTOOL_REGISTRYwiring;- adapter discovery and framework tool rendering;
- FHIR, OMOP, HL7 v2, C-CDA, OpenMRS, DHIS2, and OpenHIM handoffs;
- LangChain, Haystack, LlamaIndex, spaCy, SQL, and distributed connector wrappers.
Route elsewhere when the task is about
- Model cache planning, backend selection, or runtime export →
model-runtimes-mobile - PHI redaction policy, reversible de-identification, or date shifting →
deidentification-privacy - Clinical entity interpretation, grounding, or note-level extraction →
clinical-extraction-grounding
Core surfaces
- CLI:
openmed,openmed.cli,openmed-mcp - REST/gRPC:
openmed.service.app.create_app,openmed.service.client.OpenMedClient,openmed.service.grpc_server - MCP:
openmed.mcp.server,openmed.mcp.tool_registry.TOOL_REGISTRY - Interop registry:
openmed.interop.available_adapters,adapter_tool_definitions,to_function_tools,to_tool_use_tools - Connector families:
openmed.interop.hl7v2,openmed.interop.cda,openmed.interop.fhir_operations,openmed.interop.fhir_server,openmed.interop.openmrs,openmed.interop.omop,openmed.interop.cdm_etl,openmed.interop.duckdb_udf,openmed.interop.langchain,openmed.interop.haystack,openmed.interop.llamaindex,openmed.interop.spacy_component,openmed.interop.search_pipeline,openmed.interop.spark_udf,openmed.interop.pandas_accessor,openmed.interop.polars_accessor,openmed.interop.beam_transform,openmed.interop.ray_data,openmed.interop.prefect_tasks
Default operating rules
- Prefer one-shot CLI for finite local tasks that can complete and exit.
- Prefer REST or gRPC when you need a shared warm pool, repeated calls, async jobs, streaming, or browser clients.
- Prefer MCP when an agent or IDE needs stable tool definitions and transport negotiation.
- Keep credentials, org-unit trees, FHIR profiles, OMOP vocab snapshots, and external EHR endpoints caller-supplied.
- Keep all examples synthetic and PHI-free.
Bundled references
references/interop-serving-workflows.mdreferences/cli-reference.mdreferences/troubleshooting.mdscripts/openmed_cli_probe.py
04 同仓库更多 Skills
passnet-triton-opt
Make a single PassNet Triton kernel fast AND numerically faithful: performance model (when a replacement can win at all), block/grid/warp tuning, autotune policy, launch overhead, and per-op numeric recipes to pass the dtype baseline tolerances. Use when a pass matches and is correct but speedup below expected, or when correctness fails by small numeric margins.
passnet-solve
END-TO-END playbook for solving one PassNet sample: analyze the computation graphs, decide the optimization strategy (what to fuse, what to replace, what to leave alone), drive the iteration loop, and maximize the sample score. This is the ENTRY skill — invoke it first for any PassNet optimization task; it tells you when to use passnet-pattern-fusion, passnet-triton-opt and passnet-feedback.
passnet-skill
PassNet GPU kernel optimization via compiler passes. Design and implement Triton-based optimization passes, create pass files under ./pass_dir/, self-evaluate with pass_evaluator, and iterate to maximize GPU speedup.
passnet-pattern-fusion
Author PassNet pass files: write patterns that actually MATCH the FX graph, pick fusion regions (which ops to absorb into one Triton kernel), and structure multi-pass submissions with the mandatory shared-dispatch architecture. Use when creating or fixing pattern()/replacement_args()/replacement_func() files, when a pass "failed to match", or when deciding how to fuse multiple kernels/ops.
passnet-orchestrate
Round/batch planning and decision support for PassNet graph optimization. Use for multi-sample triage, worker allocation, eval-budget policy, and post-eval keep/revert decisions. For solving one concrete sample end to end, use passnet-solve as the entry skill; consult this skill only when you need broader planning or a specific decision gate.
passnet-feedback
Fast, deterministic feedback for PassNet work: pre-flight pattern/match verification WITHOUT burning a GPU evaluation, per-node bottleneck analysis of a sample's graphs, and eval-log parsing into per-variant status + estimated score + failure classification. Use BEFORE every GPU evaluation (check_pattern), at the START of a sample (analyze_graph), and AFTER every evaluation (parse_eval_log).