{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://raw.githubusercontent.com/bitwise-media-group/evolve/main/schemas/results.schema.json",
  "title": "evolve results file (schema 5)",
  "description": "One skill's committed results (evals/<skill>/results.<ext>; format selected by results_format). The file is nested model-major: each model key maps to that model's trigger and eval entries (skill -> model -> {triggers, evals}). Each per-eval result is a superset of a skill-creator grading.json document. Entries also retain prior states — previous (the run replaced) on both tiers and baseline (the same evals run without the skill) on the eval tier — as a trimmed results array, so deltas are computable without re-running. Consumers must treat an absent key and an explicit null identically.",
  "type": "object",
  "required": ["schema", "plugin", "skill"],
  "properties": {
    "$schema": { "type": "string" },
    "schema": { "const": 5 },
    "plugin": { "type": "string" },
    "skill": { "type": "string" },
    "models": {
      "allOf": [{ "$ref": "common.schema.json#/$defs/modelKeyMap" }],
      "additionalProperties": { "$ref": "#/$defs/modelEntry" }
    }
  },
  "additionalProperties": true,
  "$defs": {
    "modelEntry": {
      "description": "One model's trigger and eval entries; either may be absent when that tier has not run.",
      "type": "object",
      "properties": {
        "triggers": { "$ref": "#/$defs/triggerEntry" },
        "evals": { "$ref": "#/$defs/evalEntry" }
      },
      "additionalProperties": true
    },
    "triggerEntry": {
      "allOf": [
        { "$ref": "common.schema.json#/$defs/header" },
        {
          "type": "object",
          "required": ["results", "summary"],
          "properties": {
            "results": { "type": "array", "items": { "$ref": "#/$defs/triggerResult" } },
            "summary": { "$ref": "#/$defs/triggerSummary" },
            "previous": { "$ref": "#/$defs/triggerSnapshot" }
          }
        }
      ]
    },
    "triggerSnapshot": {
      "description": "A record of one prior trigger run: its summary plus a results array of the same triggerResult shape as the live run.",
      "type": "object",
      "required": ["summary"],
      "properties": {
        "ran_at": { "type": "string", "format": "date-time" },
        "summary": { "$ref": "#/$defs/triggerSummary" },
        "results": { "type": "array", "items": { "$ref": "#/$defs/triggerResult" } }
      },
      "additionalProperties": true
    },
    "triggerResult": {
      "type": "object",
      "required": ["query", "should_trigger"],
      "properties": {
        "query": { "type": "string" },
        "should_trigger": { "type": "boolean" },
        "hits": { "type": "integer", "minimum": 0 },
        "runs": { "type": "integer", "minimum": 1 },
        "passed": { "type": ["boolean", "null"] },
        "avg_run_seconds": { "type": "number", "minimum": 0 },
        "estimate": { "$ref": "common.schema.json#/$defs/estimate" },
        "spec_hash": {
          "description": "Fingerprint of this trigger's authored JSON definition when the result was written; `evolve run --modified` reruns the query when it differs. Absent on results written before fingerprinting.",
          "type": "string"
        }
      },
      "additionalProperties": true
    },
    "triggerSummary": {
      "type": "object",
      "required": ["total"],
      "properties": {
        "passed": { "type": "integer", "minimum": 0 },
        "failed": { "type": "integer", "minimum": 0 },
        "total": { "type": "integer", "minimum": 0 },
        "pass_rate": { "type": "number", "minimum": 0, "maximum": 1 },
        "avg_run_seconds": { "type": "number", "minimum": 0 },
        "estimate": { "$ref": "common.schema.json#/$defs/estimate" }
      },
      "additionalProperties": true
    },
    "evalEntry": {
      "allOf": [
        { "$ref": "common.schema.json#/$defs/header" },
        {
          "type": "object",
          "required": ["results", "summary"],
          "properties": {
            "results": { "type": "array", "items": { "$ref": "#/$defs/evalResult" } },
            "summary": { "$ref": "#/$defs/evalSummary" },
            "baseline": { "$ref": "#/$defs/evalSnapshot" },
            "previous": { "$ref": "#/$defs/evalSnapshot" }
          }
        }
      ]
    },
    "evalSnapshot": {
      "description": "A record of one prior eval run (previous or without-skill baseline): its summary plus a results array of the same evalResult shape as the live run, trimmed of the bulky expectations/execution-metrics/timing detail.",
      "type": "object",
      "required": ["summary"],
      "properties": {
        "ran_at": { "type": "string", "format": "date-time" },
        "summary": { "$ref": "#/$defs/evalSummary" },
        "results": { "type": "array", "items": { "$ref": "#/$defs/evalResult" } }
      },
      "additionalProperties": true
    },
    "evalResult": {
      "description": "One eval's outcome: grading.json's expectations/summary/execution_metrics/timing (and, once populated, claims/user_notes_summary/eval_feedback) plus evolve's identity, token, and cost extras. Count-only runs carry just id and estimate.",
      "type": "object",
      "required": ["id"],
      "properties": {
        "id": { "type": "string" },
        "name": { "type": "string" },
        "passed": { "type": ["boolean", "null"] },
        "estimate": { "$ref": "common.schema.json#/$defs/estimate" },
        "measured": { "$ref": "common.schema.json#/$defs/measured" },
        "expectations": {
          "type": "array",
          "items": { "$ref": "grading.schema.json#/$defs/expectation" }
        },
        "summary": { "$ref": "grading.schema.json#/$defs/summary" },
        "execution_metrics": { "$ref": "metrics.schema.json" },
        "timing": { "$ref": "timing.schema.json" },
        "claims": {
          "type": "array",
          "items": { "$ref": "grading.schema.json#/$defs/claim" }
        },
        "user_notes_summary": { "$ref": "grading.schema.json#/$defs/userNotesSummary" },
        "eval_feedback": { "$ref": "grading.schema.json#/$defs/evalFeedback" },
        "spec_hash": {
          "description": "Fingerprint of this eval's authored JSON definition when the result was written; `evolve run --modified` reruns the eval when it differs. Absent on results written before fingerprinting.",
          "type": "string"
        },
        "fingerprint": {
          "description": "Baseline-snapshot results only: the eval spec+fixtures fingerprint the baseline was run against, so a baseline is recomputed only when the eval changes. Absent elsewhere.",
          "type": "string"
        }
      },
      "additionalProperties": true
    },
    "evalSummary": {
      "type": "object",
      "required": ["total"],
      "properties": {
        "passed": { "type": "integer", "minimum": 0 },
        "failed": { "type": "integer", "minimum": 0 },
        "total": { "type": "integer", "minimum": 0 },
        "pass_rate": { "type": "number", "minimum": 0, "maximum": 1 },
        "avg_run_seconds": { "type": "number", "minimum": 0 },
        "estimate": { "$ref": "common.schema.json#/$defs/estimate" },
        "measured": { "$ref": "common.schema.json#/$defs/measured" }
      },
      "additionalProperties": true
    }
  }
}