Non-autoregressive System 1 decision engine. Typed choice, score and yes/no decisions over any text in a single forward pass, in 100+ languages, with a router that picks the right checkpoint per request.
Python calibrationclassificationdecision-modelhuggingfacejevmodernbertmultilingualnlppythonpytorchroutingtyped-decisionszero-shot
fix(mcp): align model schema nullability
Ha1baraA11b259749ae468
diff --git a/docs/cli-mcp.md b/docs/cli-mcp.mdindex 736436be..bcdb1f87 100644--- a/docs/cli-mcp.md+++ b/docs/cli-mcp.md@@ -166,6 +166,8 @@ Laya does not open a network port.| `laya_route_batch` | Decides which checkpoint would answer each request, with no forward pass and no checkpoint load. | `requests`, each `{state, questions, model?, task?, lang?, lang_guess?}`, optional `hooks_timeout` || `laya_decide` | Answers a JSON-schema-shaped decision in one forward pass and returns the decided values with per-field confidence, instead of an answer map to parse. Schema properties may be enum choices, booleans, or integers with a minimum and maximum; free strings, arrays, and nested objects are rejected by path. | `state`, `schema`, optional `model`, `min_confidence` |+On `laya_route`, `laya_predict`, `laya_shortlist`, and `laya_decide`, `model` may be omitted or set to `null`; either form resolves to `auto`.+The three batch and schema tools exist because the same operations are available on the SDK and`laya-serve`: handling many requests, or serving a caller that already knows the answer shape,does not require dropping to Python. For the schema-driven form in more depth, seediff --git a/laya/mcp/server.py b/laya/mcp/server.pyindex 41a57990..89b56cac 100644--- a/laya/mcp/server.py+++ b/laya/mcp/server.py@@ -293,7 +293,7 @@ def laya_route_tool(def laya_predict_tool(state: dict,questions: dict,- model: str = "auto",+ model: str | None = None,task: str | None = None,lang: str | None = None,lang_guess: str | None = None,@@ -397,7 +397,7 @@ def laya_route_batch_tool(requests: list, hooks_timeout: float = 0) -> str:def laya_shortlist_tool(state: dict,questions: dict,- model: str = "auto",+ model: str | None = None,k: int = 20,task: str | None = None,lang: str | None = None,@@ -498,7 +498,7 @@ def laya_preset_tool(+ _GUARDRAILS),)-def laya_decide_tool(state: dict, schema: dict, model: str = "auto",+def laya_decide_tool(state: dict, schema: dict, model: str | None = None,min_confidence: float | None = None) -> str:"""Answer a JSON-schema-shaped decision and return the decided values."""router = _router_or_error()diff --git a/tests/test_mcp.py b/tests/test_mcp.pyindex a0611990..5ded2a28 100644--- a/tests/test_mcp.py+++ b/tests/test_mcp.py@@ -1831,6 +1831,21 @@ def test_controls_signature_and_schema():required == (["preset", "state"] if name == "laya_preset" else ["state", "questions"]),repr(required))+ from laya.mcp import server as server_mod++ model_schemas = []+ for name in ("laya_predict", "laya_shortlist", "laya_decide", "laya_route"):+ parameter = inspect.signature(getattr(server_mod, name + "_tool")).parameters["model"]+ ok("signature/%s_model_default_none" % name, parameter.default is None, repr(parameter.default))+ tool = by_name[name]+ schema = tool.input_schema if hasattr(tool, "input_schema") else tool.inputSchema+ model_schema = schema.get("properties", {}).get("model", {})+ ok("schema/%s_model_nullable" % name,+ {"type": "null"} in (model_schema.get("anyOf") or []), repr(model_schema))+ model_schemas.append(model_schema)+ ok("schema/model_specs_consistent", all(model_schema == model_schemas[0] for model_schema in model_schemas),+ repr(model_schemas))+def test_lang_guess_control():"""`lang_guess` -- core's soft routing hint (#489 family) -- reaches the single-request tools.