@thinkai/tai-api-contract 2.175.0 → 2.176.0

package/src/generated/openapi.ts

@@ -4894,6 +4894,50 @@ export interface paths {
         patch?: never;
         trace?: never;
     };
+    "/workspaces/{workspaceId}/evals/suites/{suiteId}/compare": {
+        parameters: {
+            query?: never;
+            header?: never;
+            path?: never;
+            cookie?: never;
+        };
+        /**
+         * Compare two succeeded eval runs on a suite
+         * @description Side-by-side compare of a baseline run vs a candidate run on the same suite. Both runs must be `succeeded`. Aligns cases by `caseId` and reports suite- and case-level score/token/cost deltas when present. Requires workspace membership. Does not trigger scoring.
+         */
+        get: operations["compareEvalRuns"];
+        put?: never;
+        post?: never;
+        delete?: never;
+        options?: never;
+        head?: never;
+        patch?: never;
+        trace?: never;
+    };
+    "/workspaces/{workspaceId}/evals/suites/{suiteId}/decisions": {
+        parameters: {
+            query?: never;
+            header?: never;
+            path?: never;
+            cookie?: never;
+        };
+        /**
+         * List eval promote/keep/reject decisions for a suite
+         * @description Returns the last 50 decisions for the suite (newest first). Empty → `{ "decisions": [] }`. Requires workspace membership.
+         */
+        get: operations["listEvalDecisions"];
+        put?: never;
+        /**
+         * Record a promote, keep, or reject decision
+         * @description Persists an audit decision for a baseline vs candidate run pair. Requires workspace editor or admin. Both runs must be `succeeded` on this suite. `promote` also applies the candidate run's `suiteSnapshot.target` to the live suite (version may bump). Promotion gate attach is deferred (`gateApplied: false`, `gateDetail.reason: deferred`) until thinkai-platform-workspace#492.
+         */
+        post: operations["createEvalDecision"];
+        delete?: never;
+        options?: never;
+        head?: never;
+        patch?: never;
+        trace?: never;
+    };
     "/workspaces/{workspaceId}/evals/suites/{suiteId}/runs": {
         parameters: {
             query?: never;
@@ -11292,6 +11336,8 @@ export interface components {
             createdByUserId: string;
             target: components["schemas"]["EvalSuiteTargetDto"];
             cases: components["schemas"]["EvalCaseDto"][];
+            /** @description Most recent promote/keep/reject decision for this suite, when any exist. Omitted or null when none recorded. */
+            latestDecision?: components["schemas"]["EvalDecisionDto"] | null;
         };
         EvalSuiteListDto: {
             suites: components["schemas"]["EvalSuiteDto"][];
@@ -11442,6 +11488,89 @@ export interface components {
         EvalPackListDto: {
             packs: components["schemas"]["EvalPackSummaryDto"][];
         };
+        EvalCompareRunSummaryDto: {
+            /** Format: uuid */
+            id: string;
+            suiteVersion: number;
+            score: number | null;
+            /** Format: date-time */
+            finishedAt: string | null;
+            target: components["schemas"]["EvalSuiteTargetDto"];
+            passCount: number;
+            failCount: number;
+            errorCount: number;
+        };
+        EvalCompareCaseDto: {
+            caseId: string;
+            name: string;
+            baselineStatus: (string & components["schemas"]["EvalCaseResultStatusDto"]) | null;
+            candidateStatus: (string & components["schemas"]["EvalCaseResultStatusDto"]) | null;
+            baselinePass: boolean | null;
+            candidatePass: boolean | null;
+            baselineFailedSignalIds: string[];
+            candidateFailedSignalIds: string[];
+            baselineInputTokens?: number | null;
+            candidateInputTokens?: number | null;
+            baselineOutputTokens?: number | null;
+            candidateOutputTokens?: number | null;
+            baselineCostCents?: number | null;
+            candidateCostCents?: number | null;
+            /** @description Candidate pass (1/0) minus baseline pass when both scored. */
+            scoreDelta?: number | null;
+            inputTokensDelta?: number | null;
+            outputTokensDelta?: number | null;
+            costCentsDelta?: number | null;
+        };
+        EvalCompareDto: {
+            /** Format: uuid */
+            suiteId: string;
+            baseline: components["schemas"]["EvalCompareRunSummaryDto"];
+            candidate: components["schemas"]["EvalCompareRunSummaryDto"];
+            /** @description candidate.score − baseline.score when both present. */
+            scoreDelta: number | null;
+            avgInputTokensDelta?: number | null;
+            avgOutputTokensDelta?: number | null;
+            avgCostCentsDelta?: number | null;
+            cases: components["schemas"]["EvalCompareCaseDto"][];
+            /** @description Case ids present on only one run's results/snapshot. */
+            unpairedCaseIds: string[];
+        };
+        /** @enum {string} */
+        EvalDecisionKindDto: "promote" | "keep" | "reject";
+        EvalGateDetailDto: {
+            /** @description Machine-readable gate outcome. `deferred` until promotion-gate attach (thinkai-platform-workspace#492) ships. */
+            reason: string;
+        };
+        EvalDecisionDto: {
+            /** Format: uuid */
+            id: string;
+            /** Format: uuid */
+            suiteId: string;
+            decision: components["schemas"]["EvalDecisionKindDto"];
+            /** Format: uuid */
+            baselineRunId: string;
+            /** Format: uuid */
+            candidateRunId: string;
+            /** Format: uuid */
+            decidedByUserId: string;
+            /** Format: date-time */
+            decidedAt: string;
+            note?: string | null;
+            /** @description True when a promotion gate attach point enforced this decision. */
+            gateApplied: boolean;
+            gateDetail?: components["schemas"]["EvalGateDetailDto"] | null;
+        };
+        EvalDecisionListDto: {
+            decisions: components["schemas"]["EvalDecisionDto"][];
+        };
+        CreateEvalDecisionBodyDto: {
+            decision: components["schemas"]["EvalDecisionKindDto"];
+            /** Format: uuid */
+            baselineRunId: string;
+            /** Format: uuid */
+            candidateRunId: string;
+            note?: string;
+        };
     };
     responses: {
         /** @description Missing or invalid bearer token */
@@ -12160,6 +12289,14 @@ export type EvalModelComparisonDto = components['schemas']['EvalModelComparisonD
 export type EvalInsightsDto = components['schemas']['EvalInsightsDto'];
 export type EvalPackSummaryDto = components['schemas']['EvalPackSummaryDto'];
 export type EvalPackListDto = components['schemas']['EvalPackListDto'];
+export type EvalCompareRunSummaryDto = components['schemas']['EvalCompareRunSummaryDto'];
+export type EvalCompareCaseDto = components['schemas']['EvalCompareCaseDto'];
+export type EvalCompareDto = components['schemas']['EvalCompareDto'];
+export type EvalDecisionKindDto = components['schemas']['EvalDecisionKindDto'];
+export type EvalGateDetailDto = components['schemas']['EvalGateDetailDto'];
+export type EvalDecisionDto = components['schemas']['EvalDecisionDto'];
+export type EvalDecisionListDto = components['schemas']['EvalDecisionListDto'];
+export type CreateEvalDecisionBodyDto = components['schemas']['CreateEvalDecisionBodyDto'];
 export type ResponseUnauthorized = components['responses']['Unauthorized'];
 export type ResponseForbidden = components['responses']['Forbidden'];
 export type ResponseZenHrSyncInProgress = components['responses']['ZenHrSyncInProgress'];
@@ -25095,6 +25232,155 @@ export interface operations {
             };
         };
     };
+    compareEvalRuns: {
+        parameters: {
+            query: {
+                baselineRunId: string;
+                candidateRunId: string;
+            };
+            header?: never;
+            path: {
+                workspaceId: components["parameters"]["WorkspaceId"];
+                /** @description Eval suite id (`tai_workspace_eval_suites.id`). */
+                suiteId: components["parameters"]["EvalSuiteId"];
+            };
+            cookie?: never;
+        };
+        requestBody?: never;
+        responses: {
+            /** @description Compare result */
+            200: {
+                headers: {
+                    [name: string]: unknown;
+                };
+                content: {
+                    "application/json": components["schemas"]["EvalCompareDto"];
+                };
+            };
+            /** @description Runs not comparable (e.g. not succeeded, same id, or invalid query). Codes: `eval_compare_runs_required`, `eval_compare_runs_not_succeeded`, `eval_compare_same_run`. */
+            400: {
+                headers: {
+                    [name: string]: unknown;
+                };
+                content: {
+                    "application/json": components["schemas"]["ErrorMessageDto"];
+                };
+            };
+            401: components["responses"]["Unauthorized"];
+            403: components["responses"]["Forbidden"];
+            /** @description Workspace, suite, or run not found */
+            404: {
+                headers: {
+                    [name: string]: unknown;
+                };
+                content: {
+                    "application/json": components["schemas"]["ErrorMessageDto"];
+                };
+            };
+        };
+    };
+    listEvalDecisions: {
+        parameters: {
+            query?: never;
+            header?: never;
+            path: {
+                workspaceId: components["parameters"]["WorkspaceId"];
+                /** @description Eval suite id (`tai_workspace_eval_suites.id`). */
+                suiteId: components["parameters"]["EvalSuiteId"];
+            };
+            cookie?: never;
+        };
+        requestBody?: never;
+        responses: {
+            /** @description Decision history */
+            200: {
+                headers: {
+                    [name: string]: unknown;
+                };
+                content: {
+                    /**
+                     * @example {
+                     *       "decisions": []
+                     *     }
+                     */
+                    "application/json": components["schemas"]["EvalDecisionListDto"];
+                };
+            };
+            401: components["responses"]["Unauthorized"];
+            403: components["responses"]["Forbidden"];
+            /** @description Workspace or suite not found */
+            404: {
+                headers: {
+                    [name: string]: unknown;
+                };
+                content: {
+                    "application/json": components["schemas"]["ErrorMessageDto"];
+                };
+            };
+        };
+    };
+    createEvalDecision: {
+        parameters: {
+            query?: never;
+            header?: never;
+            path: {
+                workspaceId: components["parameters"]["WorkspaceId"];
+                /** @description Eval suite id (`tai_workspace_eval_suites.id`). */
+                suiteId: components["parameters"]["EvalSuiteId"];
+            };
+            cookie?: never;
+        };
+        requestBody: {
+            content: {
+                "application/json": components["schemas"]["CreateEvalDecisionBodyDto"];
+            };
+        };
+        responses: {
+            /** @description Decision recorded */
+            201: {
+                headers: {
+                    [name: string]: unknown;
+                };
+                content: {
+                    "application/json": components["schemas"]["EvalDecisionDto"];
+                };
+            };
+            /** @description Validation failed. Codes: `eval_decision_invalid`, `eval_compare_runs_not_succeeded`, `eval_compare_same_run`, `eval_promote_snapshot_missing`. */
+            400: {
+                headers: {
+                    [name: string]: unknown;
+                };
+                content: {
+                    "application/json": components["schemas"]["ErrorMessageDto"];
+                };
+            };
+            401: components["responses"]["Unauthorized"];
+            /** @description Authenticated but not a workspace editor/admin */
+            403: {
+                headers: {
+                    [name: string]: unknown;
+                };
+                content: {
+                    /**
+                     * @example {
+                     *       "error": "Workspace editor role required",
+                     *       "code": "workspace_editor_required"
+                     *     }
+                     */
+                    "application/json": components["schemas"]["ErrorMessageDto"];
+                };
+            };
+            /** @description Workspace, suite, or run not found */
+            404: {
+                headers: {
+                    [name: string]: unknown;
+                };
+                content: {
+                    "application/json": components["schemas"]["ErrorMessageDto"];
+                };
+            };
+        };
+    };
     listEvalRuns: {
         parameters: {
             query?: never;