@meetkai/mka1 0.54.23 → 0.54.25

package/dist/commonjs/mcp-server/tools/llmEvalsGetRunOverview.js

@@ -0,0 +1,64 @@
+"use strict";
+/*
+ * Code generated by Speakeasy (https://speakeasy.com). DO NOT EDIT.
+ */
+var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    var desc = Object.getOwnPropertyDescriptor(m, k);
+    if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
+      desc = { enumerable: true, get: function() { return m[k]; } };
+    }
+    Object.defineProperty(o, k2, desc);
+}) : (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    o[k2] = m[k];
+}));
+var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
+    Object.defineProperty(o, "default", { enumerable: true, value: v });
+}) : function(o, v) {
+    o["default"] = v;
+});
+var __importStar = (this && this.__importStar) || (function () {
+    var ownKeys = function(o) {
+        ownKeys = Object.getOwnPropertyNames || function (o) {
+            var ar = [];
+            for (var k in o) if (Object.prototype.hasOwnProperty.call(o, k)) ar[ar.length] = k;
+            return ar;
+        };
+        return ownKeys(o);
+    };
+    return function (mod) {
+        if (mod && mod.__esModule) return mod;
+        var result = {};
+        if (mod != null) for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
+        __setModuleDefault(result, mod);
+        return result;
+    };
+})();
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.tool$llmEvalsGetRunOverview = void 0;
+const llmEvalsGetRunOverview_js_1 = require("../../funcs/llmEvalsGetRunOverview.js");
+const operations = __importStar(require("../../models/operations/index.js"));
+const tools_js_1 = require("../tools.js");
+const args = {
+    request: operations.GetEvalRunOverviewRequest$inboundSchema,
+};
+exports.tool$llmEvalsGetRunOverview = {
+    name: "llm-evals-get-run-overview",
+    description: `Get an eval run overview
+
+Aggregate header for one run's trials view: trial counts, average reward, cost, and the input/output/cached token split. Its own route rather than fields on the run object because it scans the run's samples — folding it into \`GET /evals/runs/{run_id}\` would charge that scan to every run list and to the status polling that runs for the whole length of an eval. \`trials.errored\` overlaps \`trials.failed\` instead of partitioning it: a coding-agent trial can carry an agent timeout in \`error\` and still be completed and scored. \`trials.retried\` counts trials rerun through rerun-failed, read from the pre-rerun snapshot, since a rerun overwrites the sample in place.`,
+    args,
+    tool: async (client, args, ctx) => {
+        const [result, apiCall] = await (0, llmEvalsGetRunOverview_js_1.llmEvalsGetRunOverview)(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
+        if (!result.ok) {
+            return {
+                content: [{ type: "text", text: result.error.message }],
+                isError: true,
+            };
+        }
+        const value = result.value;
+        return (0, tools_js_1.formatResult)(value, apiCall);
+    },
+};
+//# sourceMappingURL=llmEvalsGetRunOverview.js.map

package/dist/commonjs/mcp-server/tools/llmEvalsGetRunOverview.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"llmEvalsGetRunOverview.js","sourceRoot":"","sources":["../../../../src/mcp-server/tools/llmEvalsGetRunOverview.ts"],"names":[],"mappings":";AAAA;;GAEG;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAEH,qFAA+E;AAC/E,6EAA+D;AAC/D,0CAA2D;AAE3D,MAAM,IAAI,GAAG;IACX,OAAO,EAAE,UAAU,CAAC,uCAAuC;CAC5D,CAAC;AAEW,QAAA,2BAA2B,GAAgC;IACtE,IAAI,EAAE,4BAA4B;IAClC,WAAW,EAAE;;mqBAEopB;IACjqB,IAAI;IACJ,IAAI,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,GAAG,EAAE,EAAE;QAChC,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,MAAM,IAAA,kDAAsB,EACpD,MAAM,EACN,IAAI,CAAC,OAAO,EACZ,EAAE,YAAY,EAAE,EAAE,MAAM,EAAE,GAAG,CAAC,MAAM,EAAE,EAAE,CACzC,CAAC,QAAQ,EAAE,CAAC;QAEb,IAAI,CAAC,MAAM,CAAC,EAAE,EAAE,CAAC;YACf,OAAO;gBACL,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC;gBACvD,OAAO,EAAE,IAAI;aACd,CAAC;QACJ,CAAC;QAED,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC;QAE3B,OAAO,IAAA,uBAAY,EAAC,KAAK,EAAE,OAAO,CAAC,CAAC;IACtC,CAAC;CACF,CAAC"}

package/dist/commonjs/mcp-server/tools/llmEvalsGetSampleTrajectory.d.ts

@@ -0,0 +1,8 @@
+import * as operations from "../../models/operations/index.js";
+import { ToolDefinition } from "../tools.js";
+declare const args: {
+    request: import("zod").ZodType<operations.GetEvalSampleTrajectoryRequest, import("zod").ZodTypeDef, unknown>;
+};
+export declare const tool$llmEvalsGetSampleTrajectory: ToolDefinition<typeof args>;
+export {};
+//# sourceMappingURL=llmEvalsGetSampleTrajectory.d.ts.map

package/dist/commonjs/mcp-server/tools/llmEvalsGetSampleTrajectory.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"llmEvalsGetSampleTrajectory.d.ts","sourceRoot":"","sources":["../../../../src/mcp-server/tools/llmEvalsGetSampleTrajectory.ts"],"names":[],"mappings":"AAKA,OAAO,KAAK,UAAU,MAAM,kCAAkC,CAAC;AAC/D,OAAO,EAAgB,cAAc,EAAE,MAAM,aAAa,CAAC;AAE3D,QAAA,MAAM,IAAI;;CAET,CAAC;AAEF,eAAO,MAAM,gCAAgC,EAAE,cAAc,CAAC,OAAO,IAAI,CAwBxE,CAAC"}

package/dist/commonjs/mcp-server/tools/llmEvalsGetSampleTrajectory.js

@@ -0,0 +1,64 @@
+"use strict";
+/*
+ * Code generated by Speakeasy (https://speakeasy.com). DO NOT EDIT.
+ */
+var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    var desc = Object.getOwnPropertyDescriptor(m, k);
+    if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
+      desc = { enumerable: true, get: function() { return m[k]; } };
+    }
+    Object.defineProperty(o, k2, desc);
+}) : (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    o[k2] = m[k];
+}));
+var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
+    Object.defineProperty(o, "default", { enumerable: true, value: v });
+}) : function(o, v) {
+    o["default"] = v;
+});
+var __importStar = (this && this.__importStar) || (function () {
+    var ownKeys = function(o) {
+        ownKeys = Object.getOwnPropertyNames || function (o) {
+            var ar = [];
+            for (var k in o) if (Object.prototype.hasOwnProperty.call(o, k)) ar[ar.length] = k;
+            return ar;
+        };
+        return ownKeys(o);
+    };
+    return function (mod) {
+        if (mod && mod.__esModule) return mod;
+        var result = {};
+        if (mod != null) for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
+        __setModuleDefault(result, mod);
+        return result;
+    };
+})();
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.tool$llmEvalsGetSampleTrajectory = void 0;
+const llmEvalsGetSampleTrajectory_js_1 = require("../../funcs/llmEvalsGetSampleTrajectory.js");
+const operations = __importStar(require("../../models/operations/index.js"));
+const tools_js_1 = require("../tools.js");
+const args = {
+    request: operations.GetEvalSampleTrajectoryRequest$inboundSchema,
+};
+exports.tool$llmEvalsGetSampleTrajectory = {
+    name: "llm-evals-get-sample-trajectory",
+    description: `Get an eval sample trajectory
+
+Returns one coding-agent trial's session — the steps the agent took, verbatim in the harness's interchange format (see \`schema_version\`, e.g. \`ATIF-v1.7\`). Deliberately not part of the samples list: a trajectory averages ~92 KB and that list pages up to 500 rows. Sample indices are unique per (task, model), not per run, so pass \`task_id\` and/or \`model\` when the run covers more than one; an index that matches several trials returns 400 rather than guessing, and never falls back to another trial's session. Returns 404 when the sample does not exist or recorded no trajectory.`,
+    args,
+    tool: async (client, args, ctx) => {
+        const [result, apiCall] = await (0, llmEvalsGetSampleTrajectory_js_1.llmEvalsGetSampleTrajectory)(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
+        if (!result.ok) {
+            return {
+                content: [{ type: "text", text: result.error.message }],
+                isError: true,
+            };
+        }
+        const value = result.value;
+        return (0, tools_js_1.formatResult)(value, apiCall);
+    },
+};
+//# sourceMappingURL=llmEvalsGetSampleTrajectory.js.map

package/dist/commonjs/mcp-server/tools/llmEvalsGetSampleTrajectory.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"llmEvalsGetSampleTrajectory.js","sourceRoot":"","sources":["../../../../src/mcp-server/tools/llmEvalsGetSampleTrajectory.ts"],"names":[],"mappings":";AAAA;;GAEG;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAEH,+FAAyF;AACzF,6EAA+D;AAC/D,0CAA2D;AAE3D,MAAM,IAAI,GAAG;IACX,OAAO,EAAE,UAAU,CAAC,4CAA4C;CACjE,CAAC;AAEW,QAAA,gCAAgC,GAAgC;IAC3E,IAAI,EAAE,iCAAiC;IACvC,WAAW,EAAE;;4kBAE6jB;IAC1kB,IAAI;IACJ,IAAI,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,GAAG,EAAE,EAAE;QAChC,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,MAAM,IAAA,4DAA2B,EACzD,MAAM,EACN,IAAI,CAAC,OAAO,EACZ,EAAE,YAAY,EAAE,EAAE,MAAM,EAAE,GAAG,CAAC,MAAM,EAAE,EAAE,CACzC,CAAC,QAAQ,EAAE,CAAC;QAEb,IAAI,CAAC,MAAM,CAAC,EAAE,EAAE,CAAC;YACf,OAAO;gBACL,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC;gBACvD,OAAO,EAAE,IAAI;aACd,CAAC;QACJ,CAAC;QAED,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC;QAE3B,OAAO,IAAA,uBAAY,EAAC,KAAK,EAAE,OAAO,CAAC,CAAC;IACtC,CAAC;CACF,CAAC"}

package/dist/commonjs/mcp-server/tools/llmEvalsGetSuiteLeaderboard.d.ts

@@ -0,0 +1,8 @@
+import * as operations from "../../models/operations/index.js";
+import { ToolDefinition } from "../tools.js";
+declare const args: {
+    request: import("zod").ZodType<operations.GetEvalSuiteLeaderboardRequest, import("zod").ZodTypeDef, unknown>;
+};
+export declare const tool$llmEvalsGetSuiteLeaderboard: ToolDefinition<typeof args>;
+export {};
+//# sourceMappingURL=llmEvalsGetSuiteLeaderboard.d.ts.map

package/dist/commonjs/mcp-server/tools/llmEvalsGetSuiteLeaderboard.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"llmEvalsGetSuiteLeaderboard.d.ts","sourceRoot":"","sources":["../../../../src/mcp-server/tools/llmEvalsGetSuiteLeaderboard.ts"],"names":[],"mappings":"AAKA,OAAO,KAAK,UAAU,MAAM,kCAAkC,CAAC;AAC/D,OAAO,EAAgB,cAAc,EAAE,MAAM,aAAa,CAAC;AAE3D,QAAA,MAAM,IAAI;;CAET,CAAC;AAEF,eAAO,MAAM,gCAAgC,EAAE,cAAc,CAAC,OAAO,IAAI,CAwBxE,CAAC"}

package/dist/commonjs/mcp-server/tools/llmEvalsGetSuiteLeaderboard.js

@@ -0,0 +1,64 @@
+"use strict";
+/*
+ * Code generated by Speakeasy (https://speakeasy.com). DO NOT EDIT.
+ */
+var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    var desc = Object.getOwnPropertyDescriptor(m, k);
+    if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
+      desc = { enumerable: true, get: function() { return m[k]; } };
+    }
+    Object.defineProperty(o, k2, desc);
+}) : (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    o[k2] = m[k];
+}));
+var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
+    Object.defineProperty(o, "default", { enumerable: true, value: v });
+}) : function(o, v) {
+    o["default"] = v;
+});
+var __importStar = (this && this.__importStar) || (function () {
+    var ownKeys = function(o) {
+        ownKeys = Object.getOwnPropertyNames || function (o) {
+            var ar = [];
+            for (var k in o) if (Object.prototype.hasOwnProperty.call(o, k)) ar[ar.length] = k;
+            return ar;
+        };
+        return ownKeys(o);
+    };
+    return function (mod) {
+        if (mod && mod.__esModule) return mod;
+        var result = {};
+        if (mod != null) for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
+        __setModuleDefault(result, mod);
+        return result;
+    };
+})();
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.tool$llmEvalsGetSuiteLeaderboard = void 0;
+const llmEvalsGetSuiteLeaderboard_js_1 = require("../../funcs/llmEvalsGetSuiteLeaderboard.js");
+const operations = __importStar(require("../../models/operations/index.js"));
+const tools_js_1 = require("../tools.js");
+const args = {
+    request: operations.GetEvalSuiteLeaderboardRequest$inboundSchema,
+};
+exports.tool$llmEvalsGetSuiteLeaderboard = {
+    name: "llm-evals-get-suite-leaderboard",
+    description: `Get the coding-agent leaderboard for an eval suite
+
+Ranks the suite's completed coding-agent runs, one row per (agent, model, effort). Only runs that recorded an \`agent_name\` participate — a QA or transcription run of the same suite has no agent to attribute a score to and is not a board row. \`accuracy\` is the mean of \`metric\` (default \`reward\`) over the row's scored trials, and \`accuracy_ci\` is the 95% Wald interval around it; both are derived per request rather than stored, so rerunning a failed trial moves them instead of leaving a stale copy behind. \`agent_version\` is reported but is not part of the row identity, so an agent upgrade does not split a row in two. Rows are ordered by accuracy descending, nulls last. Pass \`suite_version\` when the suite has more than one, otherwise the board mixes task sets.`,
+    args,
+    tool: async (client, args, ctx) => {
+        const [result, apiCall] = await (0, llmEvalsGetSuiteLeaderboard_js_1.llmEvalsGetSuiteLeaderboard)(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
+        if (!result.ok) {
+            return {
+                content: [{ type: "text", text: result.error.message }],
+                isError: true,
+            };
+        }
+        const value = result.value;
+        return (0, tools_js_1.formatResult)(value, apiCall);
+    },
+};
+//# sourceMappingURL=llmEvalsGetSuiteLeaderboard.js.map

package/dist/commonjs/mcp-server/tools/llmEvalsGetSuiteLeaderboard.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"llmEvalsGetSuiteLeaderboard.js","sourceRoot":"","sources":["../../../../src/mcp-server/tools/llmEvalsGetSuiteLeaderboard.ts"],"names":[],"mappings":";AAAA;;GAEG;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAEH,+FAAyF;AACzF,6EAA+D;AAC/D,0CAA2D;AAE3D,MAAM,IAAI,GAAG;IACX,OAAO,EAAE,UAAU,CAAC,4CAA4C;CACjE,CAAC;AAEW,QAAA,gCAAgC,GAAgC;IAC3E,IAAI,EAAE,iCAAiC;IACvC,WAAW,EAAE;;6wBAE8vB;IAC3wB,IAAI;IACJ,IAAI,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,GAAG,EAAE,EAAE;QAChC,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,MAAM,IAAA,4DAA2B,EACzD,MAAM,EACN,IAAI,CAAC,OAAO,EACZ,EAAE,YAAY,EAAE,EAAE,MAAM,EAAE,GAAG,CAAC,MAAM,EAAE,EAAE,CACzC,CAAC,QAAQ,EAAE,CAAC;QAEb,IAAI,CAAC,MAAM,CAAC,EAAE,EAAE,CAAC;YACf,OAAO;gBACL,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC;gBACvD,OAAO,EAAE,IAAI;aACd,CAAC;QACJ,CAAC;QAED,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC;QAE3B,OAAO,IAAA,uBAAY,EAAC,KAAK,EAAE,OAAO,CAAC,CAAC;IACtC,CAAC;CACF,CAAC"}

package/dist/commonjs/mcp-server/tools/llmEvalsImportAgentRun.d.ts

@@ -0,0 +1,8 @@
+import * as operations from "../../models/operations/index.js";
+import { ToolDefinition } from "../tools.js";
+declare const args: {
+    request: import("zod").ZodType<operations.ImportAgentEvalRunRequest, import("zod").ZodTypeDef, unknown>;
+};
+export declare const tool$llmEvalsImportAgentRun: ToolDefinition<typeof args>;
+export {};
+//# sourceMappingURL=llmEvalsImportAgentRun.d.ts.map

package/dist/commonjs/mcp-server/tools/llmEvalsImportAgentRun.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"llmEvalsImportAgentRun.d.ts","sourceRoot":"","sources":["../../../../src/mcp-server/tools/llmEvalsImportAgentRun.ts"],"names":[],"mappings":"AAKA,OAAO,KAAK,UAAU,MAAM,kCAAkC,CAAC;AAC/D,OAAO,EAAgB,cAAc,EAAE,MAAM,aAAa,CAAC;AAE3D,QAAA,MAAM,IAAI;;CAET,CAAC;AAEF,eAAO,MAAM,2BAA2B,EAAE,cAAc,CAAC,OAAO,IAAI,CAwBnE,CAAC"}

package/dist/commonjs/mcp-server/tools/llmEvalsImportAgentRun.js

@@ -0,0 +1,64 @@
+"use strict";
+/*
+ * Code generated by Speakeasy (https://speakeasy.com). DO NOT EDIT.
+ */
+var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    var desc = Object.getOwnPropertyDescriptor(m, k);
+    if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
+      desc = { enumerable: true, get: function() { return m[k]; } };
+    }
+    Object.defineProperty(o, k2, desc);
+}) : (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    o[k2] = m[k];
+}));
+var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
+    Object.defineProperty(o, "default", { enumerable: true, value: v });
+}) : function(o, v) {
+    o["default"] = v;
+});
+var __importStar = (this && this.__importStar) || (function () {
+    var ownKeys = function(o) {
+        ownKeys = Object.getOwnPropertyNames || function (o) {
+            var ar = [];
+            for (var k in o) if (Object.prototype.hasOwnProperty.call(o, k)) ar[ar.length] = k;
+            return ar;
+        };
+        return ownKeys(o);
+    };
+    return function (mod) {
+        if (mod && mod.__esModule) return mod;
+        var result = {};
+        if (mod != null) for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
+        __setModuleDefault(result, mod);
+        return result;
+    };
+})();
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.tool$llmEvalsImportAgentRun = void 0;
+const llmEvalsImportAgentRun_js_1 = require("../../funcs/llmEvalsImportAgentRun.js");
+const operations = __importStar(require("../../models/operations/index.js"));
+const tools_js_1 = require("../tools.js");
+const args = {
+    request: operations.ImportAgentEvalRunRequest$inboundSchema,
+};
+exports.tool$llmEvalsImportAgentRun = {
+    name: "llm-evals-import-agent-run",
+    description: `Import a finished coding-agent run
+
+Uploads one completed harness job — the run and its trials — directly, so results never round-trip through external storage the way the Hugging Face importer does. Every \`task_id\` must already exist in the target suite version's manifest; unknown ids are rejected rather than invented, because a typo would otherwise import as a task the suite does not have and read as a model that never covered the version. The run is recorded as \`completed\` and is not executed: this is a write, not a run request, so no workflow is started. Trials that cover every task of the version get \`task_ids: null\`, which is what lets the run rank as a complete sweep. Trajectories are NOT part of this payload — a full 91-task run measured 8.3 MB of them — so upload each one against \`PUT /evals/runs/{run_id}/samples/{sample_index}/trajectory\` afterwards. Pass \`idempotency_key\` and a retried upload returns the first run instead of doubling the trials behind a board row.`,
+    args,
+    tool: async (client, args, ctx) => {
+        const [result, apiCall] = await (0, llmEvalsImportAgentRun_js_1.llmEvalsImportAgentRun)(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
+        if (!result.ok) {
+            return {
+                content: [{ type: "text", text: result.error.message }],
+                isError: true,
+            };
+        }
+        const value = result.value;
+        return (0, tools_js_1.formatResult)(value, apiCall);
+    },
+};
+//# sourceMappingURL=llmEvalsImportAgentRun.js.map

package/dist/commonjs/mcp-server/tools/llmEvalsImportAgentRun.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"llmEvalsImportAgentRun.js","sourceRoot":"","sources":["../../../../src/mcp-server/tools/llmEvalsImportAgentRun.ts"],"names":[],"mappings":";AAAA;;GAEG;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAEH,qFAA+E;AAC/E,6EAA+D;AAC/D,0CAA2D;AAE3D,MAAM,IAAI,GAAG;IACX,OAAO,EAAE,UAAU,CAAC,uCAAuC;CAC5D,CAAC;AAEW,QAAA,2BAA2B,GAAgC;IACtE,IAAI,EAAE,4BAA4B;IAClC,WAAW,EAAE;;o8BAEq7B;IACl8B,IAAI;IACJ,IAAI,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,GAAG,EAAE,EAAE;QAChC,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,MAAM,IAAA,kDAAsB,EACpD,MAAM,EACN,IAAI,CAAC,OAAO,EACZ,EAAE,YAAY,EAAE,EAAE,MAAM,EAAE,GAAG,CAAC,MAAM,EAAE,EAAE,CACzC,CAAC,QAAQ,EAAE,CAAC;QAEb,IAAI,CAAC,MAAM,CAAC,EAAE,EAAE,CAAC;YACf,OAAO;gBACL,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC;gBACvD,OAAO,EAAE,IAAI;aACd,CAAC;QACJ,CAAC;QAED,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC;QAE3B,OAAO,IAAA,uBAAY,EAAC,KAAK,EAAE,OAAO,CAAC,CAAC;IACtC,CAAC;CACF,CAAC"}

package/dist/commonjs/mcp-server/tools/llmEvalsUploadSampleTrajectory.d.ts

@@ -0,0 +1,8 @@
+import * as operations from "../../models/operations/index.js";
+import { ToolDefinition } from "../tools.js";
+declare const args: {
+    request: import("zod").ZodType<operations.UploadEvalSampleTrajectoryRequest, import("zod").ZodTypeDef, unknown>;
+};
+export declare const tool$llmEvalsUploadSampleTrajectory: ToolDefinition<typeof args>;
+export {};
+//# sourceMappingURL=llmEvalsUploadSampleTrajectory.d.ts.map

package/dist/commonjs/mcp-server/tools/llmEvalsUploadSampleTrajectory.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"llmEvalsUploadSampleTrajectory.d.ts","sourceRoot":"","sources":["../../../../src/mcp-server/tools/llmEvalsUploadSampleTrajectory.ts"],"names":[],"mappings":"AAKA,OAAO,KAAK,UAAU,MAAM,kCAAkC,CAAC;AAC/D,OAAO,EAAgB,cAAc,EAAE,MAAM,aAAa,CAAC;AAE3D,QAAA,MAAM,IAAI;;CAET,CAAC;AAEF,eAAO,MAAM,mCAAmC,EAAE,cAAc,CAAC,OAAO,IAAI,CAyBzE,CAAC"}

package/dist/commonjs/mcp-server/tools/llmEvalsUploadSampleTrajectory.js

@@ -0,0 +1,64 @@
+"use strict";
+/*
+ * Code generated by Speakeasy (https://speakeasy.com). DO NOT EDIT.
+ */
+var __createBinding = (this && this.__createBinding) || (Object.create ? (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    var desc = Object.getOwnPropertyDescriptor(m, k);
+    if (!desc || ("get" in desc ? !m.__esModule : desc.writable || desc.configurable)) {
+      desc = { enumerable: true, get: function() { return m[k]; } };
+    }
+    Object.defineProperty(o, k2, desc);
+}) : (function(o, m, k, k2) {
+    if (k2 === undefined) k2 = k;
+    o[k2] = m[k];
+}));
+var __setModuleDefault = (this && this.__setModuleDefault) || (Object.create ? (function(o, v) {
+    Object.defineProperty(o, "default", { enumerable: true, value: v });
+}) : function(o, v) {
+    o["default"] = v;
+});
+var __importStar = (this && this.__importStar) || (function () {
+    var ownKeys = function(o) {
+        ownKeys = Object.getOwnPropertyNames || function (o) {
+            var ar = [];
+            for (var k in o) if (Object.prototype.hasOwnProperty.call(o, k)) ar[ar.length] = k;
+            return ar;
+        };
+        return ownKeys(o);
+    };
+    return function (mod) {
+        if (mod && mod.__esModule) return mod;
+        var result = {};
+        if (mod != null) for (var k = ownKeys(mod), i = 0; i < k.length; i++) if (k[i] !== "default") __createBinding(result, mod, k[i]);
+        __setModuleDefault(result, mod);
+        return result;
+    };
+})();
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.tool$llmEvalsUploadSampleTrajectory = void 0;
+const llmEvalsUploadSampleTrajectory_js_1 = require("../../funcs/llmEvalsUploadSampleTrajectory.js");
+const operations = __importStar(require("../../models/operations/index.js"));
+const tools_js_1 = require("../tools.js");
+const args = {
+    request: operations.UploadEvalSampleTrajectoryRequest$inboundSchema,
+};
+exports.tool$llmEvalsUploadSampleTrajectory = {
+    name: "llm-evals-upload-sample-trajectory",
+    description: `Upload an eval sample trajectory
+
+Stores one coding-agent trial's session, in whatever interchange format the harness produced (record it in \`schema_version\`, e.g. \`ATIF-v1.7\`). \`steps\` is stored verbatim and never validated, so a harness format change needs no gateway release. Separate from the run import because a trajectory averages ~92 KB and a whole run's worth would not fit in one request. Upserts, so re-sending after a timeout replaces the trajectory rather than failing. Sample indices are unique per (task, model), not per run, so pass \`task_id\` and/or \`model\` when the run covers more than one — an ambiguous index returns 400 rather than writing to the wrong trial. Rejects a payload whose \`steps\` and \`logs\` together exceed 16 MiB with 413.`,
+    args,
+    tool: async (client, args, ctx) => {
+        const [result, apiCall] = await (0, llmEvalsUploadSampleTrajectory_js_1.llmEvalsUploadSampleTrajectory)(client, args.request, { fetchOptions: { signal: ctx.signal } }).$inspect();
+        if (!result.ok) {
+            return {
+                content: [{ type: "text", text: result.error.message }],
+                isError: true,
+            };
+        }
+        const value = result.value;
+        return (0, tools_js_1.formatResult)(value, apiCall);
+    },
+};
+//# sourceMappingURL=llmEvalsUploadSampleTrajectory.js.map

package/dist/commonjs/mcp-server/tools/llmEvalsUploadSampleTrajectory.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"llmEvalsUploadSampleTrajectory.js","sourceRoot":"","sources":["../../../../src/mcp-server/tools/llmEvalsUploadSampleTrajectory.ts"],"names":[],"mappings":";AAAA;;GAEG;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AAEH,qGAA+F;AAC/F,6EAA+D;AAC/D,0CAA2D;AAE3D,MAAM,IAAI,GAAG;IACX,OAAO,EAAE,UAAU,CAAC,+CAA+C;CACpE,CAAC;AAEW,QAAA,mCAAmC,GAC9C;IACE,IAAI,EAAE,oCAAoC;IAC1C,WAAW,EAAE;;iuBAEgtB;IAC7tB,IAAI;IACJ,IAAI,EAAE,KAAK,EAAE,MAAM,EAAE,IAAI,EAAE,GAAG,EAAE,EAAE;QAChC,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,GAAG,MAAM,IAAA,kEAA8B,EAC5D,MAAM,EACN,IAAI,CAAC,OAAO,EACZ,EAAE,YAAY,EAAE,EAAE,MAAM,EAAE,GAAG,CAAC,MAAM,EAAE,EAAE,CACzC,CAAC,QAAQ,EAAE,CAAC;QAEb,IAAI,CAAC,MAAM,CAAC,EAAE,EAAE,CAAC;YACf,OAAO;gBACL,OAAO,EAAE,CAAC,EAAE,IAAI,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,CAAC,KAAK,CAAC,OAAO,EAAE,CAAC;gBACvD,OAAO,EAAE,IAAI;aACd,CAAC;QACJ,CAAC;QAED,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,CAAC;QAE3B,OAAO,IAAA,uBAAY,EAAC,KAAK,EAAE,OAAO,CAAC,CAAC;IACtC,CAAC;CACF,CAAC"}

package/dist/commonjs/models/components/appendagenttrialsrequest.d.ts

@@ -0,0 +1,111 @@
+import * as z from "zod/v3";
+import { ClosedEnum } from "../../types/enums.js";
+import { Result as SafeParseResult } from "../../types/fp.js";
+import { SDKValidationError } from "../errors/sdkvalidationerror.js";
+/**
+ * `failed` means no score was produced at all. An agent timeout is NOT that — the harness still runs the verifier and scores the trial, so it stays `completed` and carries its exception in `error`.
+ */
+export declare const AppendAgentTrialsRequestStatus: {
+    readonly Completed: "completed";
+    readonly Failed: "failed";
+};
+/**
+ * `failed` means no score was produced at all. An agent timeout is NOT that — the harness still runs the verifier and scores the trial, so it stays `completed` and carries its exception in `error`.
+ */
+export type AppendAgentTrialsRequestStatus = ClosedEnum<typeof AppendAgentTrialsRequestStatus>;
+export type AppendAgentTrialsRequestTrials = {
+    /**
+     * Must already exist in the target suite version's manifest; unknown ids are rejected rather than invented.
+     */
+    taskId: string;
+    sampleIndex: number;
+    /**
+     * `failed` means no score was produced at all. An agent timeout is NOT that — the harness still runs the verifier and scores the trial, so it stays `completed` and carries its exception in `error`.
+     */
+    status?: AppendAgentTrialsRequestStatus | undefined;
+    /**
+     * The task instruction.
+     */
+    prompt?: string | undefined;
+    target?: string | null | undefined;
+    /**
+     * What decided the score — the verifier's output. The agent's session goes to the trajectory route.
+     */
+    outputText?: string | null | undefined;
+    /**
+     * Metric values for this trial, e.g. `{ "reward": 1, "resolved": 1 }`. The board averages `reward` by default.
+     */
+    scores?: {
+        [k: string]: number;
+    } | undefined;
+    /**
+     * The exception this trial recorded, if any. Does not by itself make the trial a failure.
+     */
+    error?: {
+        [k: string]: any;
+    } | null | undefined;
+    costUsd?: number | null | undefined;
+    inputTokens?: number | null | undefined;
+    outputTokens?: number | null | undefined;
+    /**
+     * The share of `input_tokens` served from the provider's prompt cache.
+     */
+    cachedTokens?: number | null | undefined;
+    phaseTimings?: {
+        [k: string]: any;
+    } | null | undefined;
+};
+export type AppendAgentTrialsRequest = {
+    /**
+     * Each trial pins its own `sample_index` (unique per task and model): re-sending a batch replaces the same rows instead of duplicating them, so a retried batch is safe.
+     */
+    trials?: Array<AppendAgentTrialsRequestTrials> | undefined;
+    /**
+     * Marks the run completed once these trials land — the last call of a streaming job sends this, usually with an empty `trials`.
+     */
+    finalize?: boolean | undefined;
+};
+/** @internal */
+export declare const AppendAgentTrialsRequestStatus$inboundSchema: z.ZodNativeEnum<typeof AppendAgentTrialsRequestStatus>;
+/** @internal */
+export declare const AppendAgentTrialsRequestStatus$outboundSchema: z.ZodNativeEnum<typeof AppendAgentTrialsRequestStatus>;
+/** @internal */
+export declare const AppendAgentTrialsRequestTrials$inboundSchema: z.ZodType<AppendAgentTrialsRequestTrials, z.ZodTypeDef, unknown>;
+/** @internal */
+export type AppendAgentTrialsRequestTrials$Outbound = {
+    task_id: string;
+    sample_index: number;
+    status: string;
+    prompt: string;
+    target?: string | null | undefined;
+    output_text?: string | null | undefined;
+    scores?: {
+        [k: string]: number;
+    } | undefined;
+    error?: {
+        [k: string]: any;
+    } | null | undefined;
+    cost_usd?: number | null | undefined;
+    input_tokens?: number | null | undefined;
+    output_tokens?: number | null | undefined;
+    cached_tokens?: number | null | undefined;
+    phase_timings?: {
+        [k: string]: any;
+    } | null | undefined;
+};
+/** @internal */
+export declare const AppendAgentTrialsRequestTrials$outboundSchema: z.ZodType<AppendAgentTrialsRequestTrials$Outbound, z.ZodTypeDef, AppendAgentTrialsRequestTrials>;
+export declare function appendAgentTrialsRequestTrialsToJSON(appendAgentTrialsRequestTrials: AppendAgentTrialsRequestTrials): string;
+export declare function appendAgentTrialsRequestTrialsFromJSON(jsonString: string): SafeParseResult<AppendAgentTrialsRequestTrials, SDKValidationError>;
+/** @internal */
+export declare const AppendAgentTrialsRequest$inboundSchema: z.ZodType<AppendAgentTrialsRequest, z.ZodTypeDef, unknown>;
+/** @internal */
+export type AppendAgentTrialsRequest$Outbound = {
+    trials?: Array<AppendAgentTrialsRequestTrials$Outbound> | undefined;
+    finalize: boolean;
+};
+/** @internal */
+export declare const AppendAgentTrialsRequest$outboundSchema: z.ZodType<AppendAgentTrialsRequest$Outbound, z.ZodTypeDef, AppendAgentTrialsRequest>;
+export declare function appendAgentTrialsRequestToJSON(appendAgentTrialsRequest: AppendAgentTrialsRequest): string;
+export declare function appendAgentTrialsRequestFromJSON(jsonString: string): SafeParseResult<AppendAgentTrialsRequest, SDKValidationError>;
+//# sourceMappingURL=appendagenttrialsrequest.d.ts.map

package/dist/commonjs/models/components/appendagenttrialsrequest.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"appendagenttrialsrequest.d.ts","sourceRoot":"","sources":["../../../../src/models/components/appendagenttrialsrequest.ts"],"names":[],"mappings":"AAIA,OAAO,KAAK,CAAC,MAAM,QAAQ,CAAC;AAG5B,OAAO,EAAE,UAAU,EAAE,MAAM,sBAAsB,CAAC;AAClD,OAAO,EAAE,MAAM,IAAI,eAAe,EAAE,MAAM,mBAAmB,CAAC;AAC9D,OAAO,EAAE,kBAAkB,EAAE,MAAM,iCAAiC,CAAC;AAErE;;GAEG;AACH,eAAO,MAAM,8BAA8B;;;CAGjC,CAAC;AACX;;GAEG;AACH,MAAM,MAAM,8BAA8B,GAAG,UAAU,CACrD,OAAO,8BAA8B,CACtC,CAAC;AAEF,MAAM,MAAM,8BAA8B,GAAG;IAC3C;;OAEG;IACH,MAAM,EAAE,MAAM,CAAC;IACf,WAAW,EAAE,MAAM,CAAC;IACpB;;OAEG;IACH,MAAM,CAAC,EAAE,8BAA8B,GAAG,SAAS,CAAC;IACpD;;OAEG;IACH,MAAM,CAAC,EAAE,MAAM,GAAG,SAAS,CAAC;IAC5B,MAAM,CAAC,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,CAAC;IACnC;;OAEG;IACH,UAAU,CAAC,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,CAAC;IACvC;;OAEG;IACH,MAAM,CAAC,EAAE;QAAE,CAAC,CAAC,EAAE,MAAM,GAAG,MAAM,CAAA;KAAE,GAAG,SAAS,CAAC;IAC7C;;OAEG;IACH,KAAK,CAAC,EAAE;QAAE,CAAC,CAAC,EAAE,MAAM,GAAG,GAAG,CAAA;KAAE,GAAG,IAAI,GAAG,SAAS,CAAC;IAChD,OAAO,CAAC,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,CAAC;IACpC,WAAW,CAAC,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,CAAC;IACxC,YAAY,CAAC,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,CAAC;IACzC;;OAEG;IACH,YAAY,CAAC,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,CAAC;IACzC,YAAY,CAAC,EAAE;QAAE,CAAC,CAAC,EAAE,MAAM,GAAG,GAAG,CAAA;KAAE,GAAG,IAAI,GAAG,SAAS,CAAC;CACxD,CAAC;AAEF,MAAM,MAAM,wBAAwB,GAAG;IACrC;;OAEG;IACH,MAAM,CAAC,EAAE,KAAK,CAAC,8BAA8B,CAAC,GAAG,SAAS,CAAC;IAC3D;;OAEG;IACH,QAAQ,CAAC,EAAE,OAAO,GAAG,SAAS,CAAC;CAChC,CAAC;AAEF,gBAAgB;AAChB,eAAO,MAAM,4CAA4C,EAAE,CAAC,CAAC,aAAa,CACxE,OAAO,8BAA8B,CACS,CAAC;AACjD,gBAAgB;AAChB,eAAO,MAAM,6CAA6C,EAAE,CAAC,CAAC,aAAa,CACzE,OAAO,8BAA8B,CACS,CAAC;AAEjD,gBAAgB;AAChB,eAAO,MAAM,4CAA4C,EAAE,CAAC,CAAC,OAAO,CAClE,8BAA8B,EAC9B,CAAC,CAAC,UAAU,EACZ,OAAO,CA0BP,CAAC;AACH,gBAAgB;AAChB,MAAM,MAAM,uCAAuC,GAAG;IACpD,OAAO,EAAE,MAAM,CAAC;IAChB,YAAY,EAAE,MAAM,CAAC;IACrB,MAAM,EAAE,MAAM,CAAC;IACf,MAAM,EAAE,MAAM,CAAC;IACf,MAAM,CAAC,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,CAAC;IACnC,WAAW,CAAC,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,CAAC;IACxC,MAAM,CAAC,EAAE;QAAE,CAAC,CAAC,EAAE,MAAM,GAAG,MAAM,CAAA;KAAE,GAAG,SAAS,CAAC;IAC7C,KAAK,CAAC,EAAE;QAAE,CAAC,CAAC,EAAE,MAAM,GAAG,GAAG,CAAA;KAAE,GAAG,IAAI,GAAG,SAAS,CAAC;IAChD,QAAQ,CAAC,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,CAAC;IACrC,YAAY,CAAC,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,CAAC;IACzC,aAAa,CAAC,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,CAAC;IAC1C,aAAa,CAAC,EAAE,MAAM,GAAG,IAAI,GAAG,SAAS,CAAC;IAC1C,aAAa,CAAC,EAAE;QAAE,CAAC,CAAC,EAAE,MAAM,GAAG,GAAG,CAAA;KAAE,GAAG,IAAI,GAAG,SAAS,CAAC;CACzD,CAAC;AAEF,gBAAgB;AAChB,eAAO,MAAM,6CAA6C,EAAE,CAAC,CAAC,OAAO,CACnE,uCAAuC,EACvC,CAAC,CAAC,UAAU,EACZ,8BAA8B,CA0B9B,CAAC;AAEH,wBAAgB,oCAAoC,CAClD,8BAA8B,EAAE,8BAA8B,GAC7D,MAAM,CAMR;AACD,wBAAgB,sCAAsC,CACpD,UAAU,EAAE,MAAM,GACjB,eAAe,CAAC,8BAA8B,EAAE,kBAAkB,CAAC,CAMrE;AAED,gBAAgB;AAChB,eAAO,MAAM,sCAAsC,EAAE,CAAC,CAAC,OAAO,CAC5D,wBAAwB,EACxB,CAAC,CAAC,UAAU,EACZ,OAAO,CAKP,CAAC;AACH,gBAAgB;AAChB,MAAM,MAAM,iCAAiC,GAAG;IAC9C,MAAM,CAAC,EAAE,KAAK,CAAC,uCAAuC,CAAC,GAAG,SAAS,CAAC;IACpE,QAAQ,EAAE,OAAO,CAAC;CACnB,CAAC;AAEF,gBAAgB;AAChB,eAAO,MAAM,uCAAuC,EAAE,CAAC,CAAC,OAAO,CAC7D,iCAAiC,EACjC,CAAC,CAAC,UAAU,EACZ,wBAAwB,CAKxB,CAAC;AAEH,wBAAgB,8BAA8B,CAC5C,wBAAwB,EAAE,wBAAwB,GACjD,MAAM,CAIR;AACD,wBAAgB,gCAAgC,CAC9C,UAAU,EAAE,MAAM,GACjB,eAAe,CAAC,wBAAwB,EAAE,kBAAkB,CAAC,CAM/D"}