judgevet.adapters.inbound.mcp_output_schemas

Status: draft.

judgevet.adapters.inbound.mcp_output_schemas

MCP output schemas for the structured content of each successful tool result.

Each schema is a closed JSON Schema 2020-12 object without a $schema key, per the 2025-11-25 output schema rules. The shapes follow the structured content that judgevet.adapters.inbound.mcp_handlers and judgevet.adapters.inbound.mcp_policy build. Error results carry no structured content and fall outside these schemas. Source: https://github.com/Alberto-Codes/judgevet/issues/211#issuecomment-5860666386. The ask_score schema alone carries default_criteria, which says whether the server applied the default rubric; evaluate_policy score answers omit it. Source: https://github.com/Alberto-Codes/judgevet/issues/228#issuecomment-5882662573.

Examples:

from judgevet.adapters.inbound.mcp_output_schemas import noul_output_schema

assert noul_output_schema()["required"] == ["noul", "model", "usage"]
See Also

choice_output_schema()

Return the ask_choice output schema.

Returns:

Type Description
dict[str, Any]

Closed object schema for choice, confidence, probabilities,
dict[str, Any]

model and usage.

noul_output_schema()

Return the ask_noul output schema.

Returns:

Type Description
dict[str, Any]

Closed object schema for noul, model and usage.

policy_output_schema()

Return the evaluate_policy output schema.

Returns:

Type Description
dict[str, Any]

Closed object schema for model, usage, answers and policy.

score_output_schema()

Return the ask_score output schema.

Returns:

Type Description
dict[str, Any]

Closed object schema for score, confidence, legend,
dict[str, Any]

probabilities, default_criteria, model and usage.