@axiom-lattice/core 3.0.3 → 3.0.4
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/dist/index.d.mts +108 -94
- package/dist/index.d.ts +108 -94
- package/dist/index.js +157 -84
- package/dist/index.js.map +1 -1
- package/dist/index.mjs +157 -84
- package/dist/index.mjs.map +1 -1
- package/package.json +1 -1
package/dist/index.d.mts
CHANGED
|
@@ -5276,100 +5276,6 @@ declare function setBindingRegistry(r: BindingRegistry): void;
|
|
|
5276
5276
|
*/
|
|
5277
5277
|
declare function getBindingRegistry(): BindingRegistry;
|
|
5278
5278
|
|
|
5279
|
-
/**
|
|
5280
|
-
* Service contract for controlling evaluation runs.
|
|
5281
|
-
*
|
|
5282
|
-
* The gateway layer implements this interface and registers it via
|
|
5283
|
-
* {@link setEvalRunService} so that agent tools (e.g., `run_eval`) can
|
|
5284
|
-
* start, abort, and monitor evaluation runs without knowing the
|
|
5285
|
-
* underlying run-time infrastructure.
|
|
5286
|
-
*
|
|
5287
|
-
* @see {@link setEvalRunService}
|
|
5288
|
-
* @see {@link getEvalRunService}
|
|
5289
|
-
*/
|
|
5290
|
-
interface EvalRunService {
|
|
5291
|
-
/**
|
|
5292
|
-
* Start a new evaluation run for the given project.
|
|
5293
|
-
*
|
|
5294
|
-
* @param tenantId - Tenant that owns the project
|
|
5295
|
-
* @param projectId - Project to evaluate
|
|
5296
|
-
* @param suiteIds - Optional suite filter — only these suites run.
|
|
5297
|
-
* Omit to run all suites. Used to keep the validation suite
|
|
5298
|
-
* untouched during the fix loop (hold-out isolation).
|
|
5299
|
-
* @param caseIds - Optional case filter — only these cases run (across
|
|
5300
|
-
* the suites selected by `suiteIds`, or all suites if omitted).
|
|
5301
|
-
* Suites with no matching cases are skipped; an error is thrown if
|
|
5302
|
-
* no case matches at all.
|
|
5303
|
-
* @param runConfig - Runtime environment (workspace/project) the eval
|
|
5304
|
-
* executes in. Comes from the CALLER's runConfig, NOT from the eval
|
|
5305
|
-
* project's targetServerConfig. Test cases are environment-agnostic;
|
|
5306
|
-
* the run environment is decided at run time.
|
|
5307
|
-
* @returns The newly created run ID
|
|
5308
|
-
*/
|
|
5309
|
-
startRun(tenantId: string, projectId: string, suiteIds?: string[], caseIds?: string[], runConfig?: {
|
|
5310
|
-
workspaceId?: string;
|
|
5311
|
-
projectId?: string;
|
|
5312
|
-
}): Promise<string>;
|
|
5313
|
-
/**
|
|
5314
|
-
* Abort a running evaluation.
|
|
5315
|
-
*
|
|
5316
|
-
* @param runId - The run to abort
|
|
5317
|
-
* @returns `true` if the run was successfully aborted
|
|
5318
|
-
*/
|
|
5319
|
-
abortRun(runId: string): Promise<boolean>;
|
|
5320
|
-
/**
|
|
5321
|
-
* Check whether a run's runner process is still alive.
|
|
5322
|
-
*
|
|
5323
|
-
* @param runId - The run to check
|
|
5324
|
-
* @returns `true` if the runner process is still active
|
|
5325
|
-
*/
|
|
5326
|
-
isRunning(runId: string): boolean;
|
|
5327
|
-
}
|
|
5328
|
-
/**
|
|
5329
|
-
* Sets the global {@link EvalRunService} instance used by agent evaluation tools.
|
|
5330
|
-
*
|
|
5331
|
-
* The service enables the `run_eval` tool to start, abort, and monitor evaluation
|
|
5332
|
-
* runs. This must be called **at gateway startup** before any agent invocation that
|
|
5333
|
-
* expects `run_eval` to be functional.
|
|
5334
|
-
*
|
|
5335
|
-
* @example
|
|
5336
|
-
* ```ts
|
|
5337
|
-
* import { setEvalRunService } from "@axiom-lattice/core";
|
|
5338
|
-
*
|
|
5339
|
-
* const evalSvc: EvalRunService = {
|
|
5340
|
-
* startRun: async (tid, pid) => { ... },
|
|
5341
|
-
* abortRun: async (rid) => { ... },
|
|
5342
|
-
* isRunning: (rid) => { ... },
|
|
5343
|
-
* };
|
|
5344
|
-
* setEvalRunService(evalSvc);
|
|
5345
|
-
* ```
|
|
5346
|
-
*
|
|
5347
|
-
* @param s - An {@link EvalRunService} implementation
|
|
5348
|
-
* @see {@link getEvalRunService}
|
|
5349
|
-
*/
|
|
5350
|
-
declare function setEvalRunService(s: EvalRunService): void;
|
|
5351
|
-
/**
|
|
5352
|
-
* Returns the globally registered {@link EvalRunService}.
|
|
5353
|
-
*
|
|
5354
|
-
* Must be called **after** {@link setEvalRunService}. Used by the `run_eval`
|
|
5355
|
-
* tool to start, abort, and monitor evaluation runs.
|
|
5356
|
-
*
|
|
5357
|
-
* @returns The active {@link EvalRunService} instance
|
|
5358
|
-
* @throws {Error} If the service has not been initialized
|
|
5359
|
-
* @see {@link setEvalRunService}
|
|
5360
|
-
*/
|
|
5361
|
-
declare function getEvalRunService(): EvalRunService;
|
|
5362
|
-
/**
|
|
5363
|
-
* Resets the global {@link EvalRunService} to `null`.
|
|
5364
|
-
*
|
|
5365
|
-
* Restricted to the `test` environment for test isolation only. Production
|
|
5366
|
-
* code must never call this function.
|
|
5367
|
-
*
|
|
5368
|
-
* @throws {Error} If `NODE_ENV` is not `"test"`
|
|
5369
|
-
* @see {@link setEvalRunService}
|
|
5370
|
-
*/
|
|
5371
|
-
declare function clearEvalRunService(): void;
|
|
5372
|
-
|
|
5373
5279
|
interface LatticeAgentStepConfig {
|
|
5374
5280
|
agent_id: string;
|
|
5375
5281
|
override_message?: string;
|
|
@@ -5508,6 +5414,114 @@ interface CaseRunResult {
|
|
|
5508
5414
|
logs: LatticeEvalLogEvent[];
|
|
5509
5415
|
}
|
|
5510
5416
|
|
|
5417
|
+
/**
|
|
5418
|
+
* Service contract for controlling evaluation runs.
|
|
5419
|
+
*
|
|
5420
|
+
* The gateway layer implements this interface and registers it via
|
|
5421
|
+
* {@link setEvalRunService} so that agent tools (e.g., `run_eval`) can
|
|
5422
|
+
* start, abort, and monitor evaluation runs without knowing the
|
|
5423
|
+
* underlying run-time infrastructure.
|
|
5424
|
+
*
|
|
5425
|
+
* @see {@link setEvalRunService}
|
|
5426
|
+
* @see {@link getEvalRunService}
|
|
5427
|
+
*/
|
|
5428
|
+
interface EvalRunService {
|
|
5429
|
+
/**
|
|
5430
|
+
* Start a new evaluation run for the given project.
|
|
5431
|
+
*
|
|
5432
|
+
* @param tenantId - Tenant that owns the project
|
|
5433
|
+
* @param projectId - Project to evaluate
|
|
5434
|
+
* @param suiteIds - Optional suite filter — only these suites run.
|
|
5435
|
+
* Omit to run all suites. Used to keep the validation suite
|
|
5436
|
+
* untouched during the fix loop (hold-out isolation).
|
|
5437
|
+
* @param caseIds - Optional case filter — only these cases run (across
|
|
5438
|
+
* the suites selected by `suiteIds`, or all suites if omitted).
|
|
5439
|
+
* Suites with no matching cases are skipped; an error is thrown if
|
|
5440
|
+
* no case matches at all.
|
|
5441
|
+
* @param runConfig - Runtime environment (workspace/project) the eval
|
|
5442
|
+
* executes in. Comes from the CALLER's runConfig, NOT from the eval
|
|
5443
|
+
* project's targetServerConfig. Test cases are environment-agnostic;
|
|
5444
|
+
* the run environment is decided at run time.
|
|
5445
|
+
* @returns The newly created run ID
|
|
5446
|
+
*/
|
|
5447
|
+
startRun(tenantId: string, projectId: string, suiteIds?: string[], caseIds?: string[], runConfig?: {
|
|
5448
|
+
workspaceId?: string;
|
|
5449
|
+
projectId?: string;
|
|
5450
|
+
}): Promise<string>;
|
|
5451
|
+
/**
|
|
5452
|
+
* Abort a running evaluation.
|
|
5453
|
+
*
|
|
5454
|
+
* @param runId - The run to abort
|
|
5455
|
+
* @returns `true` if the run was successfully aborted
|
|
5456
|
+
*/
|
|
5457
|
+
abortRun(runId: string): Promise<boolean>;
|
|
5458
|
+
/**
|
|
5459
|
+
* Wait for a run to reach a terminal state and return its batch report.
|
|
5460
|
+
*
|
|
5461
|
+
* Resolves when the run finishes (completed, failed, or aborted) and its
|
|
5462
|
+
* status has been persisted to the store — the caller can then read the
|
|
5463
|
+
* run and its results without polling. Enables synchronous tool execution
|
|
5464
|
+
* (`run_eval start` with `wait: true`).
|
|
5465
|
+
*
|
|
5466
|
+
* @param runId - The run to wait for
|
|
5467
|
+
* @returns The final batch report of the run
|
|
5468
|
+
* @throws {Error} If the run is not tracked in this process's memory
|
|
5469
|
+
* (already finished and cleaned up, or owned by another gateway instance)
|
|
5470
|
+
*/
|
|
5471
|
+
waitForRun(runId: string): Promise<LatticeEvalBatchReport>;
|
|
5472
|
+
/**
|
|
5473
|
+
* Check whether a run's runner process is still alive.
|
|
5474
|
+
*
|
|
5475
|
+
* @param runId - The run to check
|
|
5476
|
+
* @returns `true` if the runner process is still active
|
|
5477
|
+
*/
|
|
5478
|
+
isRunning(runId: string): boolean;
|
|
5479
|
+
}
|
|
5480
|
+
/**
|
|
5481
|
+
* Sets the global {@link EvalRunService} instance used by agent evaluation tools.
|
|
5482
|
+
*
|
|
5483
|
+
* The service enables the `run_eval` tool to start, abort, and monitor evaluation
|
|
5484
|
+
* runs. This must be called **at gateway startup** before any agent invocation that
|
|
5485
|
+
* expects `run_eval` to be functional.
|
|
5486
|
+
*
|
|
5487
|
+
* @example
|
|
5488
|
+
* ```ts
|
|
5489
|
+
* import { setEvalRunService } from "@axiom-lattice/core";
|
|
5490
|
+
*
|
|
5491
|
+
* const evalSvc: EvalRunService = {
|
|
5492
|
+
* startRun: async (tid, pid) => { ... },
|
|
5493
|
+
* abortRun: async (rid) => { ... },
|
|
5494
|
+
* isRunning: (rid) => { ... },
|
|
5495
|
+
* };
|
|
5496
|
+
* setEvalRunService(evalSvc);
|
|
5497
|
+
* ```
|
|
5498
|
+
*
|
|
5499
|
+
* @param s - An {@link EvalRunService} implementation
|
|
5500
|
+
* @see {@link getEvalRunService}
|
|
5501
|
+
*/
|
|
5502
|
+
declare function setEvalRunService(s: EvalRunService): void;
|
|
5503
|
+
/**
|
|
5504
|
+
* Returns the globally registered {@link EvalRunService}.
|
|
5505
|
+
*
|
|
5506
|
+
* Must be called **after** {@link setEvalRunService}. Used by the `run_eval`
|
|
5507
|
+
* tool to start, abort, and monitor evaluation runs.
|
|
5508
|
+
*
|
|
5509
|
+
* @returns The active {@link EvalRunService} instance
|
|
5510
|
+
* @throws {Error} If the service has not been initialized
|
|
5511
|
+
* @see {@link setEvalRunService}
|
|
5512
|
+
*/
|
|
5513
|
+
declare function getEvalRunService(): EvalRunService;
|
|
5514
|
+
/**
|
|
5515
|
+
* Resets the global {@link EvalRunService} to `null`.
|
|
5516
|
+
*
|
|
5517
|
+
* Restricted to the `test` environment for test isolation only. Production
|
|
5518
|
+
* code must never call this function.
|
|
5519
|
+
*
|
|
5520
|
+
* @throws {Error} If `NODE_ENV` is not `"test"`
|
|
5521
|
+
* @see {@link setEvalRunService}
|
|
5522
|
+
*/
|
|
5523
|
+
declare function clearEvalRunService(): void;
|
|
5524
|
+
|
|
5511
5525
|
interface JudgeVerdict {
|
|
5512
5526
|
pass?: boolean;
|
|
5513
5527
|
final_score?: number;
|
package/dist/index.d.ts
CHANGED
|
@@ -5276,100 +5276,6 @@ declare function setBindingRegistry(r: BindingRegistry): void;
|
|
|
5276
5276
|
*/
|
|
5277
5277
|
declare function getBindingRegistry(): BindingRegistry;
|
|
5278
5278
|
|
|
5279
|
-
/**
|
|
5280
|
-
* Service contract for controlling evaluation runs.
|
|
5281
|
-
*
|
|
5282
|
-
* The gateway layer implements this interface and registers it via
|
|
5283
|
-
* {@link setEvalRunService} so that agent tools (e.g., `run_eval`) can
|
|
5284
|
-
* start, abort, and monitor evaluation runs without knowing the
|
|
5285
|
-
* underlying run-time infrastructure.
|
|
5286
|
-
*
|
|
5287
|
-
* @see {@link setEvalRunService}
|
|
5288
|
-
* @see {@link getEvalRunService}
|
|
5289
|
-
*/
|
|
5290
|
-
interface EvalRunService {
|
|
5291
|
-
/**
|
|
5292
|
-
* Start a new evaluation run for the given project.
|
|
5293
|
-
*
|
|
5294
|
-
* @param tenantId - Tenant that owns the project
|
|
5295
|
-
* @param projectId - Project to evaluate
|
|
5296
|
-
* @param suiteIds - Optional suite filter — only these suites run.
|
|
5297
|
-
* Omit to run all suites. Used to keep the validation suite
|
|
5298
|
-
* untouched during the fix loop (hold-out isolation).
|
|
5299
|
-
* @param caseIds - Optional case filter — only these cases run (across
|
|
5300
|
-
* the suites selected by `suiteIds`, or all suites if omitted).
|
|
5301
|
-
* Suites with no matching cases are skipped; an error is thrown if
|
|
5302
|
-
* no case matches at all.
|
|
5303
|
-
* @param runConfig - Runtime environment (workspace/project) the eval
|
|
5304
|
-
* executes in. Comes from the CALLER's runConfig, NOT from the eval
|
|
5305
|
-
* project's targetServerConfig. Test cases are environment-agnostic;
|
|
5306
|
-
* the run environment is decided at run time.
|
|
5307
|
-
* @returns The newly created run ID
|
|
5308
|
-
*/
|
|
5309
|
-
startRun(tenantId: string, projectId: string, suiteIds?: string[], caseIds?: string[], runConfig?: {
|
|
5310
|
-
workspaceId?: string;
|
|
5311
|
-
projectId?: string;
|
|
5312
|
-
}): Promise<string>;
|
|
5313
|
-
/**
|
|
5314
|
-
* Abort a running evaluation.
|
|
5315
|
-
*
|
|
5316
|
-
* @param runId - The run to abort
|
|
5317
|
-
* @returns `true` if the run was successfully aborted
|
|
5318
|
-
*/
|
|
5319
|
-
abortRun(runId: string): Promise<boolean>;
|
|
5320
|
-
/**
|
|
5321
|
-
* Check whether a run's runner process is still alive.
|
|
5322
|
-
*
|
|
5323
|
-
* @param runId - The run to check
|
|
5324
|
-
* @returns `true` if the runner process is still active
|
|
5325
|
-
*/
|
|
5326
|
-
isRunning(runId: string): boolean;
|
|
5327
|
-
}
|
|
5328
|
-
/**
|
|
5329
|
-
* Sets the global {@link EvalRunService} instance used by agent evaluation tools.
|
|
5330
|
-
*
|
|
5331
|
-
* The service enables the `run_eval` tool to start, abort, and monitor evaluation
|
|
5332
|
-
* runs. This must be called **at gateway startup** before any agent invocation that
|
|
5333
|
-
* expects `run_eval` to be functional.
|
|
5334
|
-
*
|
|
5335
|
-
* @example
|
|
5336
|
-
* ```ts
|
|
5337
|
-
* import { setEvalRunService } from "@axiom-lattice/core";
|
|
5338
|
-
*
|
|
5339
|
-
* const evalSvc: EvalRunService = {
|
|
5340
|
-
* startRun: async (tid, pid) => { ... },
|
|
5341
|
-
* abortRun: async (rid) => { ... },
|
|
5342
|
-
* isRunning: (rid) => { ... },
|
|
5343
|
-
* };
|
|
5344
|
-
* setEvalRunService(evalSvc);
|
|
5345
|
-
* ```
|
|
5346
|
-
*
|
|
5347
|
-
* @param s - An {@link EvalRunService} implementation
|
|
5348
|
-
* @see {@link getEvalRunService}
|
|
5349
|
-
*/
|
|
5350
|
-
declare function setEvalRunService(s: EvalRunService): void;
|
|
5351
|
-
/**
|
|
5352
|
-
* Returns the globally registered {@link EvalRunService}.
|
|
5353
|
-
*
|
|
5354
|
-
* Must be called **after** {@link setEvalRunService}. Used by the `run_eval`
|
|
5355
|
-
* tool to start, abort, and monitor evaluation runs.
|
|
5356
|
-
*
|
|
5357
|
-
* @returns The active {@link EvalRunService} instance
|
|
5358
|
-
* @throws {Error} If the service has not been initialized
|
|
5359
|
-
* @see {@link setEvalRunService}
|
|
5360
|
-
*/
|
|
5361
|
-
declare function getEvalRunService(): EvalRunService;
|
|
5362
|
-
/**
|
|
5363
|
-
* Resets the global {@link EvalRunService} to `null`.
|
|
5364
|
-
*
|
|
5365
|
-
* Restricted to the `test` environment for test isolation only. Production
|
|
5366
|
-
* code must never call this function.
|
|
5367
|
-
*
|
|
5368
|
-
* @throws {Error} If `NODE_ENV` is not `"test"`
|
|
5369
|
-
* @see {@link setEvalRunService}
|
|
5370
|
-
*/
|
|
5371
|
-
declare function clearEvalRunService(): void;
|
|
5372
|
-
|
|
5373
5279
|
interface LatticeAgentStepConfig {
|
|
5374
5280
|
agent_id: string;
|
|
5375
5281
|
override_message?: string;
|
|
@@ -5508,6 +5414,114 @@ interface CaseRunResult {
|
|
|
5508
5414
|
logs: LatticeEvalLogEvent[];
|
|
5509
5415
|
}
|
|
5510
5416
|
|
|
5417
|
+
/**
|
|
5418
|
+
* Service contract for controlling evaluation runs.
|
|
5419
|
+
*
|
|
5420
|
+
* The gateway layer implements this interface and registers it via
|
|
5421
|
+
* {@link setEvalRunService} so that agent tools (e.g., `run_eval`) can
|
|
5422
|
+
* start, abort, and monitor evaluation runs without knowing the
|
|
5423
|
+
* underlying run-time infrastructure.
|
|
5424
|
+
*
|
|
5425
|
+
* @see {@link setEvalRunService}
|
|
5426
|
+
* @see {@link getEvalRunService}
|
|
5427
|
+
*/
|
|
5428
|
+
interface EvalRunService {
|
|
5429
|
+
/**
|
|
5430
|
+
* Start a new evaluation run for the given project.
|
|
5431
|
+
*
|
|
5432
|
+
* @param tenantId - Tenant that owns the project
|
|
5433
|
+
* @param projectId - Project to evaluate
|
|
5434
|
+
* @param suiteIds - Optional suite filter — only these suites run.
|
|
5435
|
+
* Omit to run all suites. Used to keep the validation suite
|
|
5436
|
+
* untouched during the fix loop (hold-out isolation).
|
|
5437
|
+
* @param caseIds - Optional case filter — only these cases run (across
|
|
5438
|
+
* the suites selected by `suiteIds`, or all suites if omitted).
|
|
5439
|
+
* Suites with no matching cases are skipped; an error is thrown if
|
|
5440
|
+
* no case matches at all.
|
|
5441
|
+
* @param runConfig - Runtime environment (workspace/project) the eval
|
|
5442
|
+
* executes in. Comes from the CALLER's runConfig, NOT from the eval
|
|
5443
|
+
* project's targetServerConfig. Test cases are environment-agnostic;
|
|
5444
|
+
* the run environment is decided at run time.
|
|
5445
|
+
* @returns The newly created run ID
|
|
5446
|
+
*/
|
|
5447
|
+
startRun(tenantId: string, projectId: string, suiteIds?: string[], caseIds?: string[], runConfig?: {
|
|
5448
|
+
workspaceId?: string;
|
|
5449
|
+
projectId?: string;
|
|
5450
|
+
}): Promise<string>;
|
|
5451
|
+
/**
|
|
5452
|
+
* Abort a running evaluation.
|
|
5453
|
+
*
|
|
5454
|
+
* @param runId - The run to abort
|
|
5455
|
+
* @returns `true` if the run was successfully aborted
|
|
5456
|
+
*/
|
|
5457
|
+
abortRun(runId: string): Promise<boolean>;
|
|
5458
|
+
/**
|
|
5459
|
+
* Wait for a run to reach a terminal state and return its batch report.
|
|
5460
|
+
*
|
|
5461
|
+
* Resolves when the run finishes (completed, failed, or aborted) and its
|
|
5462
|
+
* status has been persisted to the store — the caller can then read the
|
|
5463
|
+
* run and its results without polling. Enables synchronous tool execution
|
|
5464
|
+
* (`run_eval start` with `wait: true`).
|
|
5465
|
+
*
|
|
5466
|
+
* @param runId - The run to wait for
|
|
5467
|
+
* @returns The final batch report of the run
|
|
5468
|
+
* @throws {Error} If the run is not tracked in this process's memory
|
|
5469
|
+
* (already finished and cleaned up, or owned by another gateway instance)
|
|
5470
|
+
*/
|
|
5471
|
+
waitForRun(runId: string): Promise<LatticeEvalBatchReport>;
|
|
5472
|
+
/**
|
|
5473
|
+
* Check whether a run's runner process is still alive.
|
|
5474
|
+
*
|
|
5475
|
+
* @param runId - The run to check
|
|
5476
|
+
* @returns `true` if the runner process is still active
|
|
5477
|
+
*/
|
|
5478
|
+
isRunning(runId: string): boolean;
|
|
5479
|
+
}
|
|
5480
|
+
/**
|
|
5481
|
+
* Sets the global {@link EvalRunService} instance used by agent evaluation tools.
|
|
5482
|
+
*
|
|
5483
|
+
* The service enables the `run_eval` tool to start, abort, and monitor evaluation
|
|
5484
|
+
* runs. This must be called **at gateway startup** before any agent invocation that
|
|
5485
|
+
* expects `run_eval` to be functional.
|
|
5486
|
+
*
|
|
5487
|
+
* @example
|
|
5488
|
+
* ```ts
|
|
5489
|
+
* import { setEvalRunService } from "@axiom-lattice/core";
|
|
5490
|
+
*
|
|
5491
|
+
* const evalSvc: EvalRunService = {
|
|
5492
|
+
* startRun: async (tid, pid) => { ... },
|
|
5493
|
+
* abortRun: async (rid) => { ... },
|
|
5494
|
+
* isRunning: (rid) => { ... },
|
|
5495
|
+
* };
|
|
5496
|
+
* setEvalRunService(evalSvc);
|
|
5497
|
+
* ```
|
|
5498
|
+
*
|
|
5499
|
+
* @param s - An {@link EvalRunService} implementation
|
|
5500
|
+
* @see {@link getEvalRunService}
|
|
5501
|
+
*/
|
|
5502
|
+
declare function setEvalRunService(s: EvalRunService): void;
|
|
5503
|
+
/**
|
|
5504
|
+
* Returns the globally registered {@link EvalRunService}.
|
|
5505
|
+
*
|
|
5506
|
+
* Must be called **after** {@link setEvalRunService}. Used by the `run_eval`
|
|
5507
|
+
* tool to start, abort, and monitor evaluation runs.
|
|
5508
|
+
*
|
|
5509
|
+
* @returns The active {@link EvalRunService} instance
|
|
5510
|
+
* @throws {Error} If the service has not been initialized
|
|
5511
|
+
* @see {@link setEvalRunService}
|
|
5512
|
+
*/
|
|
5513
|
+
declare function getEvalRunService(): EvalRunService;
|
|
5514
|
+
/**
|
|
5515
|
+
* Resets the global {@link EvalRunService} to `null`.
|
|
5516
|
+
*
|
|
5517
|
+
* Restricted to the `test` environment for test isolation only. Production
|
|
5518
|
+
* code must never call this function.
|
|
5519
|
+
*
|
|
5520
|
+
* @throws {Error} If `NODE_ENV` is not `"test"`
|
|
5521
|
+
* @see {@link setEvalRunService}
|
|
5522
|
+
*/
|
|
5523
|
+
declare function clearEvalRunService(): void;
|
|
5524
|
+
|
|
5511
5525
|
interface JudgeVerdict {
|
|
5512
5526
|
pass?: boolean;
|
|
5513
5527
|
final_score?: number;
|