@agentv/core 4.31.0-next.1 → 4.31.1-next.1

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.cts CHANGED
@@ -1989,8 +1989,10 @@ type ExecutionDefaults = {
1989
1989
  readonly pool_slots?: number;
1990
1990
  };
1991
1991
  type ResultsConfig = {
1992
+ readonly mode: 'github';
1992
1993
  readonly repo: string;
1993
- readonly path: string;
1994
+ /** Local filesystem path for the results clone. Optional; defaults to ~/.agentv/results/<slug>/. */
1995
+ readonly path?: string;
1994
1996
  readonly auto_push?: boolean;
1995
1997
  readonly branch_prefix?: string;
1996
1998
  };
@@ -4172,7 +4174,7 @@ declare function toSnakeCaseDeep(obj: unknown): unknown;
4172
4174
  */
4173
4175
  declare function toCamelCaseDeep(obj: unknown): unknown;
4174
4176
 
4175
- interface ResultsRepoCachePaths {
4177
+ interface ResultsRepoLocalPaths {
4176
4178
  readonly rootDir: string;
4177
4179
  readonly repoDir: string;
4178
4180
  readonly statusFile: string;
@@ -4184,7 +4186,7 @@ interface ResultsRepoStatus {
4184
4186
  readonly path?: string;
4185
4187
  readonly auto_push?: boolean;
4186
4188
  readonly branch_prefix?: string;
4187
- readonly cache_dir?: string;
4189
+ readonly local_dir?: string;
4188
4190
  readonly last_synced_at?: string;
4189
4191
  readonly last_error?: string;
4190
4192
  }
@@ -4198,7 +4200,7 @@ interface PreparedResultsRepoBranch extends CheckedOutResultsRepoBranch {
4198
4200
  }
4199
4201
  declare function normalizeResultsConfig(config: ResultsConfig): Required<ResultsConfig>;
4200
4202
  declare function resolveResultsRepoUrl(repo: string): string;
4201
- declare function getResultsRepoCachePaths(repo: string): ResultsRepoCachePaths;
4203
+ declare function getResultsRepoLocalPaths(repo: string): ResultsRepoLocalPaths;
4202
4204
  declare function ensureResultsRepoClone(config: ResultsConfig): Promise<string>;
4203
4205
  declare function getResultsRepoStatus(config?: ResultsConfig): ResultsRepoStatus;
4204
4206
  declare function syncResultsRepo(config: ResultsConfig): Promise<ResultsRepoStatus>;
@@ -4227,7 +4229,7 @@ declare function createDraftResultsPr(params: {
4227
4229
  }): Promise<string>;
4228
4230
  /**
4229
4231
  * Push results directly to the base branch of the results repo.
4230
- * Handles non-fast-forward conflicts by pulling with rebase and retrying.
4232
+ * Handles non-fast-forward conflicts by fetching, rebasing, and retrying.
4231
4233
  * Returns true if artifacts were pushed, false if no changes were detected.
4232
4234
  */
4233
4235
  declare function directPushResults(params: {
@@ -4236,6 +4238,21 @@ declare function directPushResults(params: {
4236
4238
  readonly destinationPath: string;
4237
4239
  readonly commitMessage: string;
4238
4240
  }): Promise<boolean>;
4241
+ interface GitListedRun {
4242
+ run_id: string;
4243
+ experiment: string;
4244
+ timestamp: string;
4245
+ pass_rate?: number;
4246
+ target?: string;
4247
+ manifest_path: string;
4248
+ benchmark_path: string;
4249
+ display_name: string;
4250
+ test_count: number;
4251
+ avg_score: number;
4252
+ size_bytes: number;
4253
+ }
4254
+ declare function listGitRuns(repoDir: string, ref?: string): Promise<GitListedRun[]>;
4255
+ declare function materializeGitRun(repoDir: string, relativeRunPath: string, ref?: string): Promise<void>;
4239
4256
 
4240
4257
  /**
4241
4258
  * The default config directory (~/.agentv). Always resolves to the user's home
@@ -4949,4 +4966,4 @@ type AgentKernel = {
4949
4966
  };
4950
4967
  declare function createAgentKernel(): AgentKernel;
4951
4968
 
4952
- export { type AcquireWorkspaceOptions, type AgentKernel, type AgentVConfig$1 as AgentVConfig, type AgentVResolvedConfig, type AgentVConfig as AgentVTsConfig, type AgentVConfig$1 as AgentVYamlConfig, type AnthropicResolvedConfig, type ApiFormat, type ArgsMatchMode, type AssertContext, type AssertEntry, type AssertFn, type AssertResult, type AssertionEntry, type AssertionResult, type AssistantTestMessage, type AzureResolvedConfig, COMMON_TARGET_SETTINGS, type CacheConfig, type CheckedOutResultsRepoBranch, type ChildGraderResult, type ClaudeDiscoverOptions, type ClaudeResolvedConfig, type ClaudeSession, type CliResolvedConfig, CodeGrader, type CodeGraderConfig, type CodeGraderOptions, type CodexDiscoverOptions, type CodexSession, type CommandExecutor, type CompositeAggregatorConfig, CompositeGrader, type CompositeGraderConfig, type CompositeGraderOptions, type ConfidenceIntervalAggregation, type ContainsAllGraderConfig, type ContainsAnyGraderConfig, type ContainsGraderConfig, type Content, type ContentFile, type ContentImage, type ContentPreprocessorConfig, type ContentText, type ConversationAggregation, type ConversationMode, type ConversationTurn, type ConversationTurnInput, type CopilotCliResolvedConfig, type DiscoverOptions as CopilotDiscoverOptions, type CopilotLogResolvedConfig, type CopilotSdkResolvedConfig, type CopilotSession, type CopilotSessionMeta, CostGrader, type CostGraderConfig, type CostGraderOptions, type CreateContainerOptions, DEFAULT_CATEGORY, DEFAULT_EVAL_PATTERNS, DEFAULT_EXPLORATION_TOOLS, DEFAULT_GRADER_TEMPLATE, DEFAULT_THRESHOLD, type DependencyFailurePolicy, type DependencyResult, type DepsScanResult, DeterministicAssertionGrader, type DockerWorkspaceConfig, DockerWorkspaceProvider, type EndsWithGraderConfig, type EnsureSubagentsOptions, type EnsureSubagentsResult, type EnvLookup, type EqualsGraderConfig, type EvalAssertionInput, type EvalCase, type EvalConfig, type EvalMetadata, type EvalRunResult, type EvalSuiteResult, type EvalSummary, type EvalTargetRef, type EvalTest, type EvalTestInput, type EvalsJsonCase, type EvalsJsonFile, type EvaluationCache, type EvaluationContext, type EvaluationResult, type EvaluationScore, type EvaluationVerdict, type ExecInContainerOptions, type ExecResult, type ExecutionDefaults, type ExecutionError, type ExecutionMetrics, ExecutionMetricsGrader, type ExecutionMetricsGraderConfig, type ExecutionMetricsGraderOptions, type ExecutionStatus, type FailOnError, type FailureStage, FieldAccuracyGrader, type FieldAccuracyGraderConfig, type FieldAccuracyGraderOptions, type FieldAggregationType, type FieldConfig, type FieldMatchType, type GeminiResolvedConfig, type GenerateRubricsOptions, type Grader, type GraderConfig, type GraderDispatchContext, type GraderFactory, type GraderFactoryFn, type GraderKind, GraderRegistry, type GraderResult, type IcontainsAllGraderConfig, type IcontainsAnyGraderConfig, type IcontainsGraderConfig, type InlineAssertEvaluatorConfig, type IsJsonGraderConfig, type JsonObject, type JsonPrimitive, type JsonValue, LatencyGrader, type LatencyGraderConfig, type LatencyGraderOptions, LlmGrader, type LlmGraderConfig, type LlmGraderOptions, type LlmGraderPromptAssembly, type LocalPathValidationError, type MeanAggregation, type Message, type MockResolvedConfig, OTEL_BACKEND_PRESETS, type OpenAIResolvedConfig, type OpenRouterResolvedConfig, type OtelBackendPreset, type OtelExportOptions, OtelStreamingObserver, OtelTraceExporter, OtlpJsonFileExporter, type OutputMessage, PASS_THRESHOLD, type ParsedCopilotSession, type PassAtKAggregation, type PiCliResolvedConfig, type PiCodingAgentResolvedConfig, type PoolSlot, type PreparedResultsRepoBranch, type ProgressEvent, type ProjectEntry, type ProjectRegistry, type ProjectSource, type PromptInputs, type PromptScriptConfig, type Provider, type ProviderFactoryFn, type ProviderKind, ProviderRegistry, type ProviderRequest, type ProviderResponse, type ProviderStreamCallbacks, type ProviderTokenUsage, type RegexGraderConfig, type RepoCheckout, type RepoClone, type RepoConfig, type RepoDep, RepoManager, type RepoSource, type ResolvedTarget, type ResolvedWorkspaceTemplate, ResponseCache, type ResultsConfig, type ResultsRepoCachePaths, type ResultsRepoStatus, type RubricItem, type RubricsEvaluatorConfig, RunBudgetTracker, type RunEvalCaseOptions, type RunEvaluationOptions, type ScoreRange, type ScriptExecutionContext, SkillTriggerGrader, type SkillTriggerGraderConfig, type StartsWithGraderConfig, type SystemTestMessage, TEST_MESSAGE_ROLES, type TargetAccessConfig, type TargetDefinition, type TargetHooksConfig, TemplateNotDirectoryError, TemplateNotFoundError, type TestMessage, type TestMessageContent, type TestMessageRole, type TokenUsage, TokenUsageGrader, type TokenUsageGraderConfig, type TokenUsageGraderOptions, type ToolCall, type ToolTestMessage, type ToolTrajectoryExpectedItem, ToolTrajectoryGrader, type ToolTrajectoryGraderConfig, type ToolTrajectoryGraderOptions, type TraceComputeResult, type TraceSummary, type TranscriptEntry, type TranscriptJsonLine, TranscriptProvider, type TranscriptReplayEntry, type TranscriptSource, type TranspileResult, type TrialAggregation, type TrialResult, type TrialStrategy, type TrialsConfig, type TsEvalResult, type TurnFailurePolicy, type UserTestMessage, type VSCodeResolvedConfig, type WorkspaceConfig, WorkspaceCreationError, type WorkspaceEnvConfig, type WorkspaceHookConfig, type WorkspaceHooksConfig, WorkspacePoolManager, type WorkspaceScriptConfig, addProject, assembleLlmGraderPrompt, avgToolDurationMs, buildDirectoryChain, buildOutputSchema, buildPromptInputs, buildRubricOutputSchema, buildScoreRangeOutputSchema, buildSearchRoots, calculateRubricScore, captureFileChanges, checkoutResultsRepoBranch, clampScore, cleanupEvalWorkspaces, cleanupWorkspace, commitAndPushResultsBranch, computeTraceSummary, computeWorkspaceFingerprint, consumeClaudeLogEntries, consumeCodexLogEntries, consumeCopilotCliLogEntries, consumeCopilotSdkLogEntries, consumePiLogEntries, createAgentKernel, createBuiltinProviderRegistry, createBuiltinRegistry, createDraftResultsPr, createProvider, createTempWorkspace, deepEqual, defineConfig, deriveCategory, deriveProjectId, detectFormat, directPushResults, directorySizeBytes, discoverAssertions, discoverClaudeSessions, discoverCodexSessions, discoverCopilotSessions, discoverGraders, discoverProjects, discoverProviders, ensureResultsRepoClone, ensureVSCodeSubagents, evaluate, executeScript, executeWorkspaceScript, explorationRatio, extractCacheConfig, extractFailOnError, extractImageBlocks, extractJsonBlob, extractLastAssistantContent, extractTargetFromSuite, extractTargetRefsFromSuite, extractTargetsFromSuite, extractTargetsFromTestCase, extractThreshold, extractTrialsConfig, extractWorkersFromSuite, fileExists, findGitRoot, formatToolCalls, freeformEvaluationSchema, generateRubrics, getAgentvConfigDir, getAgentvHome, getOutputFilenames, getProject, getProjectsRegistryPath, getResultsRepoCachePaths, getResultsRepoStatus, getSubagentsRoot, getTextContent, getTraceStateRoot, getWorkspacePath, getWorkspacePoolRoot, getWorkspacesRoot, groupTranscriptJsonLines, initializeBaseline, isAgentSkillsFormat, isContent, isContentArray, isGraderKind, isJsonObject, isJsonValue, isNonEmptyString, isTestMessage, isTestMessageRole, killAllTrackedChildren, listTargetNames, loadConfig, loadEvalCaseById, loadEvalCases, loadEvalSuite, loadProjectRegistry, loadTestById, loadTestSuite, loadTests, loadTsConfig, loadTsEvalFile, mergeExecutionMetrics, negateScore, normalizeLineEndings, normalizeResultsConfig, parseAgentSkillsEvals, parseClaudeSession, parseCodexSession, parseCopilotEvents, parseEnvOutput, parseJsonFromText, parseJsonSafe, parseYamlValue, prepareResultsRepoBranch, pushResultsRepoBranch, readJsonFile, readTargetDefinitions, readTestSuiteMetadata, readTextFile, readTranscriptFile, readTranscriptJsonl, removeProject, resolveAndCreateProvider, resolveDelegatedTargetDefinition, resolveFileReference, resolveResultsRepoRunsDir, resolveResultsRepoUrl, resolveTargetDefinition, resolveWorkspaceTemplate, rubricEvaluationSchema, runBeforeSessionHook, runContainsAllAssertion, runContainsAnyAssertion, runContainsAssertion, runEndsWithAssertion, runEqualsAssertion, runEvalCase, runEvaluation, runIcontainsAllAssertion, runIcontainsAnyAssertion, runIcontainsAssertion, runIsJsonAssertion, runRegexAssertion, runStartsWithAssertion, saveProjectRegistry, scanRepoDeps, scoreRangeEvaluationSchema, scoreToVerdict, shouldEnableCache, shouldSkipCacheForTemperature, stageResultsArtifacts, subscribeToClaudeLogEntries, subscribeToCodexLogEntries, subscribeToCopilotCliLogEntries, subscribeToCopilotSdkLogEntries, subscribeToPiLogEntries, substituteVariables, syncProject, syncProjects, syncResultsRepo, toCamelCaseDeep, toSnakeCaseDeep, toTranscriptJsonLines, tokensPerTool, touchProject, trackChild, trackedChildCount, transpileEvalYaml, transpileEvalYamlFile, trimBaselineResult };
4969
+ export { type AcquireWorkspaceOptions, type AgentKernel, type AgentVConfig$1 as AgentVConfig, type AgentVResolvedConfig, type AgentVConfig as AgentVTsConfig, type AgentVConfig$1 as AgentVYamlConfig, type AnthropicResolvedConfig, type ApiFormat, type ArgsMatchMode, type AssertContext, type AssertEntry, type AssertFn, type AssertResult, type AssertionEntry, type AssertionResult, type AssistantTestMessage, type AzureResolvedConfig, COMMON_TARGET_SETTINGS, type CacheConfig, type CheckedOutResultsRepoBranch, type ChildGraderResult, type ClaudeDiscoverOptions, type ClaudeResolvedConfig, type ClaudeSession, type CliResolvedConfig, CodeGrader, type CodeGraderConfig, type CodeGraderOptions, type CodexDiscoverOptions, type CodexSession, type CommandExecutor, type CompositeAggregatorConfig, CompositeGrader, type CompositeGraderConfig, type CompositeGraderOptions, type ConfidenceIntervalAggregation, type ContainsAllGraderConfig, type ContainsAnyGraderConfig, type ContainsGraderConfig, type Content, type ContentFile, type ContentImage, type ContentPreprocessorConfig, type ContentText, type ConversationAggregation, type ConversationMode, type ConversationTurn, type ConversationTurnInput, type CopilotCliResolvedConfig, type DiscoverOptions as CopilotDiscoverOptions, type CopilotLogResolvedConfig, type CopilotSdkResolvedConfig, type CopilotSession, type CopilotSessionMeta, CostGrader, type CostGraderConfig, type CostGraderOptions, type CreateContainerOptions, DEFAULT_CATEGORY, DEFAULT_EVAL_PATTERNS, DEFAULT_EXPLORATION_TOOLS, DEFAULT_GRADER_TEMPLATE, DEFAULT_THRESHOLD, type DependencyFailurePolicy, type DependencyResult, type DepsScanResult, DeterministicAssertionGrader, type DockerWorkspaceConfig, DockerWorkspaceProvider, type EndsWithGraderConfig, type EnsureSubagentsOptions, type EnsureSubagentsResult, type EnvLookup, type EqualsGraderConfig, type EvalAssertionInput, type EvalCase, type EvalConfig, type EvalMetadata, type EvalRunResult, type EvalSuiteResult, type EvalSummary, type EvalTargetRef, type EvalTest, type EvalTestInput, type EvalsJsonCase, type EvalsJsonFile, type EvaluationCache, type EvaluationContext, type EvaluationResult, type EvaluationScore, type EvaluationVerdict, type ExecInContainerOptions, type ExecResult, type ExecutionDefaults, type ExecutionError, type ExecutionMetrics, ExecutionMetricsGrader, type ExecutionMetricsGraderConfig, type ExecutionMetricsGraderOptions, type ExecutionStatus, type FailOnError, type FailureStage, FieldAccuracyGrader, type FieldAccuracyGraderConfig, type FieldAccuracyGraderOptions, type FieldAggregationType, type FieldConfig, type FieldMatchType, type GeminiResolvedConfig, type GenerateRubricsOptions, type GitListedRun, type Grader, type GraderConfig, type GraderDispatchContext, type GraderFactory, type GraderFactoryFn, type GraderKind, GraderRegistry, type GraderResult, type IcontainsAllGraderConfig, type IcontainsAnyGraderConfig, type IcontainsGraderConfig, type InlineAssertEvaluatorConfig, type IsJsonGraderConfig, type JsonObject, type JsonPrimitive, type JsonValue, LatencyGrader, type LatencyGraderConfig, type LatencyGraderOptions, LlmGrader, type LlmGraderConfig, type LlmGraderOptions, type LlmGraderPromptAssembly, type LocalPathValidationError, type MeanAggregation, type Message, type MockResolvedConfig, OTEL_BACKEND_PRESETS, type OpenAIResolvedConfig, type OpenRouterResolvedConfig, type OtelBackendPreset, type OtelExportOptions, OtelStreamingObserver, OtelTraceExporter, OtlpJsonFileExporter, type OutputMessage, PASS_THRESHOLD, type ParsedCopilotSession, type PassAtKAggregation, type PiCliResolvedConfig, type PiCodingAgentResolvedConfig, type PoolSlot, type PreparedResultsRepoBranch, type ProgressEvent, type ProjectEntry, type ProjectRegistry, type ProjectSource, type PromptInputs, type PromptScriptConfig, type Provider, type ProviderFactoryFn, type ProviderKind, ProviderRegistry, type ProviderRequest, type ProviderResponse, type ProviderStreamCallbacks, type ProviderTokenUsage, type RegexGraderConfig, type RepoCheckout, type RepoClone, type RepoConfig, type RepoDep, RepoManager, type RepoSource, type ResolvedTarget, type ResolvedWorkspaceTemplate, ResponseCache, type ResultsConfig, type ResultsRepoLocalPaths, type ResultsRepoStatus, type RubricItem, type RubricsEvaluatorConfig, RunBudgetTracker, type RunEvalCaseOptions, type RunEvaluationOptions, type ScoreRange, type ScriptExecutionContext, SkillTriggerGrader, type SkillTriggerGraderConfig, type StartsWithGraderConfig, type SystemTestMessage, TEST_MESSAGE_ROLES, type TargetAccessConfig, type TargetDefinition, type TargetHooksConfig, TemplateNotDirectoryError, TemplateNotFoundError, type TestMessage, type TestMessageContent, type TestMessageRole, type TokenUsage, TokenUsageGrader, type TokenUsageGraderConfig, type TokenUsageGraderOptions, type ToolCall, type ToolTestMessage, type ToolTrajectoryExpectedItem, ToolTrajectoryGrader, type ToolTrajectoryGraderConfig, type ToolTrajectoryGraderOptions, type TraceComputeResult, type TraceSummary, type TranscriptEntry, type TranscriptJsonLine, TranscriptProvider, type TranscriptReplayEntry, type TranscriptSource, type TranspileResult, type TrialAggregation, type TrialResult, type TrialStrategy, type TrialsConfig, type TsEvalResult, type TurnFailurePolicy, type UserTestMessage, type VSCodeResolvedConfig, type WorkspaceConfig, WorkspaceCreationError, type WorkspaceEnvConfig, type WorkspaceHookConfig, type WorkspaceHooksConfig, WorkspacePoolManager, type WorkspaceScriptConfig, addProject, assembleLlmGraderPrompt, avgToolDurationMs, buildDirectoryChain, buildOutputSchema, buildPromptInputs, buildRubricOutputSchema, buildScoreRangeOutputSchema, buildSearchRoots, calculateRubricScore, captureFileChanges, checkoutResultsRepoBranch, clampScore, cleanupEvalWorkspaces, cleanupWorkspace, commitAndPushResultsBranch, computeTraceSummary, computeWorkspaceFingerprint, consumeClaudeLogEntries, consumeCodexLogEntries, consumeCopilotCliLogEntries, consumeCopilotSdkLogEntries, consumePiLogEntries, createAgentKernel, createBuiltinProviderRegistry, createBuiltinRegistry, createDraftResultsPr, createProvider, createTempWorkspace, deepEqual, defineConfig, deriveCategory, deriveProjectId, detectFormat, directPushResults, directorySizeBytes, discoverAssertions, discoverClaudeSessions, discoverCodexSessions, discoverCopilotSessions, discoverGraders, discoverProjects, discoverProviders, ensureResultsRepoClone, ensureVSCodeSubagents, evaluate, executeScript, executeWorkspaceScript, explorationRatio, extractCacheConfig, extractFailOnError, extractImageBlocks, extractJsonBlob, extractLastAssistantContent, extractTargetFromSuite, extractTargetRefsFromSuite, extractTargetsFromSuite, extractTargetsFromTestCase, extractThreshold, extractTrialsConfig, extractWorkersFromSuite, fileExists, findGitRoot, formatToolCalls, freeformEvaluationSchema, generateRubrics, getAgentvConfigDir, getAgentvHome, getOutputFilenames, getProject, getProjectsRegistryPath, getResultsRepoLocalPaths, getResultsRepoStatus, getSubagentsRoot, getTextContent, getTraceStateRoot, getWorkspacePath, getWorkspacePoolRoot, getWorkspacesRoot, groupTranscriptJsonLines, initializeBaseline, isAgentSkillsFormat, isContent, isContentArray, isGraderKind, isJsonObject, isJsonValue, isNonEmptyString, isTestMessage, isTestMessageRole, killAllTrackedChildren, listGitRuns, listTargetNames, loadConfig, loadEvalCaseById, loadEvalCases, loadEvalSuite, loadProjectRegistry, loadTestById, loadTestSuite, loadTests, loadTsConfig, loadTsEvalFile, materializeGitRun, mergeExecutionMetrics, negateScore, normalizeLineEndings, normalizeResultsConfig, parseAgentSkillsEvals, parseClaudeSession, parseCodexSession, parseCopilotEvents, parseEnvOutput, parseJsonFromText, parseJsonSafe, parseYamlValue, prepareResultsRepoBranch, pushResultsRepoBranch, readJsonFile, readTargetDefinitions, readTestSuiteMetadata, readTextFile, readTranscriptFile, readTranscriptJsonl, removeProject, resolveAndCreateProvider, resolveDelegatedTargetDefinition, resolveFileReference, resolveResultsRepoRunsDir, resolveResultsRepoUrl, resolveTargetDefinition, resolveWorkspaceTemplate, rubricEvaluationSchema, runBeforeSessionHook, runContainsAllAssertion, runContainsAnyAssertion, runContainsAssertion, runEndsWithAssertion, runEqualsAssertion, runEvalCase, runEvaluation, runIcontainsAllAssertion, runIcontainsAnyAssertion, runIcontainsAssertion, runIsJsonAssertion, runRegexAssertion, runStartsWithAssertion, saveProjectRegistry, scanRepoDeps, scoreRangeEvaluationSchema, scoreToVerdict, shouldEnableCache, shouldSkipCacheForTemperature, stageResultsArtifacts, subscribeToClaudeLogEntries, subscribeToCodexLogEntries, subscribeToCopilotCliLogEntries, subscribeToCopilotSdkLogEntries, subscribeToPiLogEntries, substituteVariables, syncProject, syncProjects, syncResultsRepo, toCamelCaseDeep, toSnakeCaseDeep, toTranscriptJsonLines, tokensPerTool, touchProject, trackChild, trackedChildCount, transpileEvalYaml, transpileEvalYamlFile, trimBaselineResult };
package/dist/index.d.ts CHANGED
@@ -1989,8 +1989,10 @@ type ExecutionDefaults = {
1989
1989
  readonly pool_slots?: number;
1990
1990
  };
1991
1991
  type ResultsConfig = {
1992
+ readonly mode: 'github';
1992
1993
  readonly repo: string;
1993
- readonly path: string;
1994
+ /** Local filesystem path for the results clone. Optional; defaults to ~/.agentv/results/<slug>/. */
1995
+ readonly path?: string;
1994
1996
  readonly auto_push?: boolean;
1995
1997
  readonly branch_prefix?: string;
1996
1998
  };
@@ -4172,7 +4174,7 @@ declare function toSnakeCaseDeep(obj: unknown): unknown;
4172
4174
  */
4173
4175
  declare function toCamelCaseDeep(obj: unknown): unknown;
4174
4176
 
4175
- interface ResultsRepoCachePaths {
4177
+ interface ResultsRepoLocalPaths {
4176
4178
  readonly rootDir: string;
4177
4179
  readonly repoDir: string;
4178
4180
  readonly statusFile: string;
@@ -4184,7 +4186,7 @@ interface ResultsRepoStatus {
4184
4186
  readonly path?: string;
4185
4187
  readonly auto_push?: boolean;
4186
4188
  readonly branch_prefix?: string;
4187
- readonly cache_dir?: string;
4189
+ readonly local_dir?: string;
4188
4190
  readonly last_synced_at?: string;
4189
4191
  readonly last_error?: string;
4190
4192
  }
@@ -4198,7 +4200,7 @@ interface PreparedResultsRepoBranch extends CheckedOutResultsRepoBranch {
4198
4200
  }
4199
4201
  declare function normalizeResultsConfig(config: ResultsConfig): Required<ResultsConfig>;
4200
4202
  declare function resolveResultsRepoUrl(repo: string): string;
4201
- declare function getResultsRepoCachePaths(repo: string): ResultsRepoCachePaths;
4203
+ declare function getResultsRepoLocalPaths(repo: string): ResultsRepoLocalPaths;
4202
4204
  declare function ensureResultsRepoClone(config: ResultsConfig): Promise<string>;
4203
4205
  declare function getResultsRepoStatus(config?: ResultsConfig): ResultsRepoStatus;
4204
4206
  declare function syncResultsRepo(config: ResultsConfig): Promise<ResultsRepoStatus>;
@@ -4227,7 +4229,7 @@ declare function createDraftResultsPr(params: {
4227
4229
  }): Promise<string>;
4228
4230
  /**
4229
4231
  * Push results directly to the base branch of the results repo.
4230
- * Handles non-fast-forward conflicts by pulling with rebase and retrying.
4232
+ * Handles non-fast-forward conflicts by fetching, rebasing, and retrying.
4231
4233
  * Returns true if artifacts were pushed, false if no changes were detected.
4232
4234
  */
4233
4235
  declare function directPushResults(params: {
@@ -4236,6 +4238,21 @@ declare function directPushResults(params: {
4236
4238
  readonly destinationPath: string;
4237
4239
  readonly commitMessage: string;
4238
4240
  }): Promise<boolean>;
4241
+ interface GitListedRun {
4242
+ run_id: string;
4243
+ experiment: string;
4244
+ timestamp: string;
4245
+ pass_rate?: number;
4246
+ target?: string;
4247
+ manifest_path: string;
4248
+ benchmark_path: string;
4249
+ display_name: string;
4250
+ test_count: number;
4251
+ avg_score: number;
4252
+ size_bytes: number;
4253
+ }
4254
+ declare function listGitRuns(repoDir: string, ref?: string): Promise<GitListedRun[]>;
4255
+ declare function materializeGitRun(repoDir: string, relativeRunPath: string, ref?: string): Promise<void>;
4239
4256
 
4240
4257
  /**
4241
4258
  * The default config directory (~/.agentv). Always resolves to the user's home
@@ -4949,4 +4966,4 @@ type AgentKernel = {
4949
4966
  };
4950
4967
  declare function createAgentKernel(): AgentKernel;
4951
4968
 
4952
- export { type AcquireWorkspaceOptions, type AgentKernel, type AgentVConfig$1 as AgentVConfig, type AgentVResolvedConfig, type AgentVConfig as AgentVTsConfig, type AgentVConfig$1 as AgentVYamlConfig, type AnthropicResolvedConfig, type ApiFormat, type ArgsMatchMode, type AssertContext, type AssertEntry, type AssertFn, type AssertResult, type AssertionEntry, type AssertionResult, type AssistantTestMessage, type AzureResolvedConfig, COMMON_TARGET_SETTINGS, type CacheConfig, type CheckedOutResultsRepoBranch, type ChildGraderResult, type ClaudeDiscoverOptions, type ClaudeResolvedConfig, type ClaudeSession, type CliResolvedConfig, CodeGrader, type CodeGraderConfig, type CodeGraderOptions, type CodexDiscoverOptions, type CodexSession, type CommandExecutor, type CompositeAggregatorConfig, CompositeGrader, type CompositeGraderConfig, type CompositeGraderOptions, type ConfidenceIntervalAggregation, type ContainsAllGraderConfig, type ContainsAnyGraderConfig, type ContainsGraderConfig, type Content, type ContentFile, type ContentImage, type ContentPreprocessorConfig, type ContentText, type ConversationAggregation, type ConversationMode, type ConversationTurn, type ConversationTurnInput, type CopilotCliResolvedConfig, type DiscoverOptions as CopilotDiscoverOptions, type CopilotLogResolvedConfig, type CopilotSdkResolvedConfig, type CopilotSession, type CopilotSessionMeta, CostGrader, type CostGraderConfig, type CostGraderOptions, type CreateContainerOptions, DEFAULT_CATEGORY, DEFAULT_EVAL_PATTERNS, DEFAULT_EXPLORATION_TOOLS, DEFAULT_GRADER_TEMPLATE, DEFAULT_THRESHOLD, type DependencyFailurePolicy, type DependencyResult, type DepsScanResult, DeterministicAssertionGrader, type DockerWorkspaceConfig, DockerWorkspaceProvider, type EndsWithGraderConfig, type EnsureSubagentsOptions, type EnsureSubagentsResult, type EnvLookup, type EqualsGraderConfig, type EvalAssertionInput, type EvalCase, type EvalConfig, type EvalMetadata, type EvalRunResult, type EvalSuiteResult, type EvalSummary, type EvalTargetRef, type EvalTest, type EvalTestInput, type EvalsJsonCase, type EvalsJsonFile, type EvaluationCache, type EvaluationContext, type EvaluationResult, type EvaluationScore, type EvaluationVerdict, type ExecInContainerOptions, type ExecResult, type ExecutionDefaults, type ExecutionError, type ExecutionMetrics, ExecutionMetricsGrader, type ExecutionMetricsGraderConfig, type ExecutionMetricsGraderOptions, type ExecutionStatus, type FailOnError, type FailureStage, FieldAccuracyGrader, type FieldAccuracyGraderConfig, type FieldAccuracyGraderOptions, type FieldAggregationType, type FieldConfig, type FieldMatchType, type GeminiResolvedConfig, type GenerateRubricsOptions, type Grader, type GraderConfig, type GraderDispatchContext, type GraderFactory, type GraderFactoryFn, type GraderKind, GraderRegistry, type GraderResult, type IcontainsAllGraderConfig, type IcontainsAnyGraderConfig, type IcontainsGraderConfig, type InlineAssertEvaluatorConfig, type IsJsonGraderConfig, type JsonObject, type JsonPrimitive, type JsonValue, LatencyGrader, type LatencyGraderConfig, type LatencyGraderOptions, LlmGrader, type LlmGraderConfig, type LlmGraderOptions, type LlmGraderPromptAssembly, type LocalPathValidationError, type MeanAggregation, type Message, type MockResolvedConfig, OTEL_BACKEND_PRESETS, type OpenAIResolvedConfig, type OpenRouterResolvedConfig, type OtelBackendPreset, type OtelExportOptions, OtelStreamingObserver, OtelTraceExporter, OtlpJsonFileExporter, type OutputMessage, PASS_THRESHOLD, type ParsedCopilotSession, type PassAtKAggregation, type PiCliResolvedConfig, type PiCodingAgentResolvedConfig, type PoolSlot, type PreparedResultsRepoBranch, type ProgressEvent, type ProjectEntry, type ProjectRegistry, type ProjectSource, type PromptInputs, type PromptScriptConfig, type Provider, type ProviderFactoryFn, type ProviderKind, ProviderRegistry, type ProviderRequest, type ProviderResponse, type ProviderStreamCallbacks, type ProviderTokenUsage, type RegexGraderConfig, type RepoCheckout, type RepoClone, type RepoConfig, type RepoDep, RepoManager, type RepoSource, type ResolvedTarget, type ResolvedWorkspaceTemplate, ResponseCache, type ResultsConfig, type ResultsRepoCachePaths, type ResultsRepoStatus, type RubricItem, type RubricsEvaluatorConfig, RunBudgetTracker, type RunEvalCaseOptions, type RunEvaluationOptions, type ScoreRange, type ScriptExecutionContext, SkillTriggerGrader, type SkillTriggerGraderConfig, type StartsWithGraderConfig, type SystemTestMessage, TEST_MESSAGE_ROLES, type TargetAccessConfig, type TargetDefinition, type TargetHooksConfig, TemplateNotDirectoryError, TemplateNotFoundError, type TestMessage, type TestMessageContent, type TestMessageRole, type TokenUsage, TokenUsageGrader, type TokenUsageGraderConfig, type TokenUsageGraderOptions, type ToolCall, type ToolTestMessage, type ToolTrajectoryExpectedItem, ToolTrajectoryGrader, type ToolTrajectoryGraderConfig, type ToolTrajectoryGraderOptions, type TraceComputeResult, type TraceSummary, type TranscriptEntry, type TranscriptJsonLine, TranscriptProvider, type TranscriptReplayEntry, type TranscriptSource, type TranspileResult, type TrialAggregation, type TrialResult, type TrialStrategy, type TrialsConfig, type TsEvalResult, type TurnFailurePolicy, type UserTestMessage, type VSCodeResolvedConfig, type WorkspaceConfig, WorkspaceCreationError, type WorkspaceEnvConfig, type WorkspaceHookConfig, type WorkspaceHooksConfig, WorkspacePoolManager, type WorkspaceScriptConfig, addProject, assembleLlmGraderPrompt, avgToolDurationMs, buildDirectoryChain, buildOutputSchema, buildPromptInputs, buildRubricOutputSchema, buildScoreRangeOutputSchema, buildSearchRoots, calculateRubricScore, captureFileChanges, checkoutResultsRepoBranch, clampScore, cleanupEvalWorkspaces, cleanupWorkspace, commitAndPushResultsBranch, computeTraceSummary, computeWorkspaceFingerprint, consumeClaudeLogEntries, consumeCodexLogEntries, consumeCopilotCliLogEntries, consumeCopilotSdkLogEntries, consumePiLogEntries, createAgentKernel, createBuiltinProviderRegistry, createBuiltinRegistry, createDraftResultsPr, createProvider, createTempWorkspace, deepEqual, defineConfig, deriveCategory, deriveProjectId, detectFormat, directPushResults, directorySizeBytes, discoverAssertions, discoverClaudeSessions, discoverCodexSessions, discoverCopilotSessions, discoverGraders, discoverProjects, discoverProviders, ensureResultsRepoClone, ensureVSCodeSubagents, evaluate, executeScript, executeWorkspaceScript, explorationRatio, extractCacheConfig, extractFailOnError, extractImageBlocks, extractJsonBlob, extractLastAssistantContent, extractTargetFromSuite, extractTargetRefsFromSuite, extractTargetsFromSuite, extractTargetsFromTestCase, extractThreshold, extractTrialsConfig, extractWorkersFromSuite, fileExists, findGitRoot, formatToolCalls, freeformEvaluationSchema, generateRubrics, getAgentvConfigDir, getAgentvHome, getOutputFilenames, getProject, getProjectsRegistryPath, getResultsRepoCachePaths, getResultsRepoStatus, getSubagentsRoot, getTextContent, getTraceStateRoot, getWorkspacePath, getWorkspacePoolRoot, getWorkspacesRoot, groupTranscriptJsonLines, initializeBaseline, isAgentSkillsFormat, isContent, isContentArray, isGraderKind, isJsonObject, isJsonValue, isNonEmptyString, isTestMessage, isTestMessageRole, killAllTrackedChildren, listTargetNames, loadConfig, loadEvalCaseById, loadEvalCases, loadEvalSuite, loadProjectRegistry, loadTestById, loadTestSuite, loadTests, loadTsConfig, loadTsEvalFile, mergeExecutionMetrics, negateScore, normalizeLineEndings, normalizeResultsConfig, parseAgentSkillsEvals, parseClaudeSession, parseCodexSession, parseCopilotEvents, parseEnvOutput, parseJsonFromText, parseJsonSafe, parseYamlValue, prepareResultsRepoBranch, pushResultsRepoBranch, readJsonFile, readTargetDefinitions, readTestSuiteMetadata, readTextFile, readTranscriptFile, readTranscriptJsonl, removeProject, resolveAndCreateProvider, resolveDelegatedTargetDefinition, resolveFileReference, resolveResultsRepoRunsDir, resolveResultsRepoUrl, resolveTargetDefinition, resolveWorkspaceTemplate, rubricEvaluationSchema, runBeforeSessionHook, runContainsAllAssertion, runContainsAnyAssertion, runContainsAssertion, runEndsWithAssertion, runEqualsAssertion, runEvalCase, runEvaluation, runIcontainsAllAssertion, runIcontainsAnyAssertion, runIcontainsAssertion, runIsJsonAssertion, runRegexAssertion, runStartsWithAssertion, saveProjectRegistry, scanRepoDeps, scoreRangeEvaluationSchema, scoreToVerdict, shouldEnableCache, shouldSkipCacheForTemperature, stageResultsArtifacts, subscribeToClaudeLogEntries, subscribeToCodexLogEntries, subscribeToCopilotCliLogEntries, subscribeToCopilotSdkLogEntries, subscribeToPiLogEntries, substituteVariables, syncProject, syncProjects, syncResultsRepo, toCamelCaseDeep, toSnakeCaseDeep, toTranscriptJsonLines, tokensPerTool, touchProject, trackChild, trackedChildCount, transpileEvalYaml, transpileEvalYamlFile, trimBaselineResult };
4969
+ export { type AcquireWorkspaceOptions, type AgentKernel, type AgentVConfig$1 as AgentVConfig, type AgentVResolvedConfig, type AgentVConfig as AgentVTsConfig, type AgentVConfig$1 as AgentVYamlConfig, type AnthropicResolvedConfig, type ApiFormat, type ArgsMatchMode, type AssertContext, type AssertEntry, type AssertFn, type AssertResult, type AssertionEntry, type AssertionResult, type AssistantTestMessage, type AzureResolvedConfig, COMMON_TARGET_SETTINGS, type CacheConfig, type CheckedOutResultsRepoBranch, type ChildGraderResult, type ClaudeDiscoverOptions, type ClaudeResolvedConfig, type ClaudeSession, type CliResolvedConfig, CodeGrader, type CodeGraderConfig, type CodeGraderOptions, type CodexDiscoverOptions, type CodexSession, type CommandExecutor, type CompositeAggregatorConfig, CompositeGrader, type CompositeGraderConfig, type CompositeGraderOptions, type ConfidenceIntervalAggregation, type ContainsAllGraderConfig, type ContainsAnyGraderConfig, type ContainsGraderConfig, type Content, type ContentFile, type ContentImage, type ContentPreprocessorConfig, type ContentText, type ConversationAggregation, type ConversationMode, type ConversationTurn, type ConversationTurnInput, type CopilotCliResolvedConfig, type DiscoverOptions as CopilotDiscoverOptions, type CopilotLogResolvedConfig, type CopilotSdkResolvedConfig, type CopilotSession, type CopilotSessionMeta, CostGrader, type CostGraderConfig, type CostGraderOptions, type CreateContainerOptions, DEFAULT_CATEGORY, DEFAULT_EVAL_PATTERNS, DEFAULT_EXPLORATION_TOOLS, DEFAULT_GRADER_TEMPLATE, DEFAULT_THRESHOLD, type DependencyFailurePolicy, type DependencyResult, type DepsScanResult, DeterministicAssertionGrader, type DockerWorkspaceConfig, DockerWorkspaceProvider, type EndsWithGraderConfig, type EnsureSubagentsOptions, type EnsureSubagentsResult, type EnvLookup, type EqualsGraderConfig, type EvalAssertionInput, type EvalCase, type EvalConfig, type EvalMetadata, type EvalRunResult, type EvalSuiteResult, type EvalSummary, type EvalTargetRef, type EvalTest, type EvalTestInput, type EvalsJsonCase, type EvalsJsonFile, type EvaluationCache, type EvaluationContext, type EvaluationResult, type EvaluationScore, type EvaluationVerdict, type ExecInContainerOptions, type ExecResult, type ExecutionDefaults, type ExecutionError, type ExecutionMetrics, ExecutionMetricsGrader, type ExecutionMetricsGraderConfig, type ExecutionMetricsGraderOptions, type ExecutionStatus, type FailOnError, type FailureStage, FieldAccuracyGrader, type FieldAccuracyGraderConfig, type FieldAccuracyGraderOptions, type FieldAggregationType, type FieldConfig, type FieldMatchType, type GeminiResolvedConfig, type GenerateRubricsOptions, type GitListedRun, type Grader, type GraderConfig, type GraderDispatchContext, type GraderFactory, type GraderFactoryFn, type GraderKind, GraderRegistry, type GraderResult, type IcontainsAllGraderConfig, type IcontainsAnyGraderConfig, type IcontainsGraderConfig, type InlineAssertEvaluatorConfig, type IsJsonGraderConfig, type JsonObject, type JsonPrimitive, type JsonValue, LatencyGrader, type LatencyGraderConfig, type LatencyGraderOptions, LlmGrader, type LlmGraderConfig, type LlmGraderOptions, type LlmGraderPromptAssembly, type LocalPathValidationError, type MeanAggregation, type Message, type MockResolvedConfig, OTEL_BACKEND_PRESETS, type OpenAIResolvedConfig, type OpenRouterResolvedConfig, type OtelBackendPreset, type OtelExportOptions, OtelStreamingObserver, OtelTraceExporter, OtlpJsonFileExporter, type OutputMessage, PASS_THRESHOLD, type ParsedCopilotSession, type PassAtKAggregation, type PiCliResolvedConfig, type PiCodingAgentResolvedConfig, type PoolSlot, type PreparedResultsRepoBranch, type ProgressEvent, type ProjectEntry, type ProjectRegistry, type ProjectSource, type PromptInputs, type PromptScriptConfig, type Provider, type ProviderFactoryFn, type ProviderKind, ProviderRegistry, type ProviderRequest, type ProviderResponse, type ProviderStreamCallbacks, type ProviderTokenUsage, type RegexGraderConfig, type RepoCheckout, type RepoClone, type RepoConfig, type RepoDep, RepoManager, type RepoSource, type ResolvedTarget, type ResolvedWorkspaceTemplate, ResponseCache, type ResultsConfig, type ResultsRepoLocalPaths, type ResultsRepoStatus, type RubricItem, type RubricsEvaluatorConfig, RunBudgetTracker, type RunEvalCaseOptions, type RunEvaluationOptions, type ScoreRange, type ScriptExecutionContext, SkillTriggerGrader, type SkillTriggerGraderConfig, type StartsWithGraderConfig, type SystemTestMessage, TEST_MESSAGE_ROLES, type TargetAccessConfig, type TargetDefinition, type TargetHooksConfig, TemplateNotDirectoryError, TemplateNotFoundError, type TestMessage, type TestMessageContent, type TestMessageRole, type TokenUsage, TokenUsageGrader, type TokenUsageGraderConfig, type TokenUsageGraderOptions, type ToolCall, type ToolTestMessage, type ToolTrajectoryExpectedItem, ToolTrajectoryGrader, type ToolTrajectoryGraderConfig, type ToolTrajectoryGraderOptions, type TraceComputeResult, type TraceSummary, type TranscriptEntry, type TranscriptJsonLine, TranscriptProvider, type TranscriptReplayEntry, type TranscriptSource, type TranspileResult, type TrialAggregation, type TrialResult, type TrialStrategy, type TrialsConfig, type TsEvalResult, type TurnFailurePolicy, type UserTestMessage, type VSCodeResolvedConfig, type WorkspaceConfig, WorkspaceCreationError, type WorkspaceEnvConfig, type WorkspaceHookConfig, type WorkspaceHooksConfig, WorkspacePoolManager, type WorkspaceScriptConfig, addProject, assembleLlmGraderPrompt, avgToolDurationMs, buildDirectoryChain, buildOutputSchema, buildPromptInputs, buildRubricOutputSchema, buildScoreRangeOutputSchema, buildSearchRoots, calculateRubricScore, captureFileChanges, checkoutResultsRepoBranch, clampScore, cleanupEvalWorkspaces, cleanupWorkspace, commitAndPushResultsBranch, computeTraceSummary, computeWorkspaceFingerprint, consumeClaudeLogEntries, consumeCodexLogEntries, consumeCopilotCliLogEntries, consumeCopilotSdkLogEntries, consumePiLogEntries, createAgentKernel, createBuiltinProviderRegistry, createBuiltinRegistry, createDraftResultsPr, createProvider, createTempWorkspace, deepEqual, defineConfig, deriveCategory, deriveProjectId, detectFormat, directPushResults, directorySizeBytes, discoverAssertions, discoverClaudeSessions, discoverCodexSessions, discoverCopilotSessions, discoverGraders, discoverProjects, discoverProviders, ensureResultsRepoClone, ensureVSCodeSubagents, evaluate, executeScript, executeWorkspaceScript, explorationRatio, extractCacheConfig, extractFailOnError, extractImageBlocks, extractJsonBlob, extractLastAssistantContent, extractTargetFromSuite, extractTargetRefsFromSuite, extractTargetsFromSuite, extractTargetsFromTestCase, extractThreshold, extractTrialsConfig, extractWorkersFromSuite, fileExists, findGitRoot, formatToolCalls, freeformEvaluationSchema, generateRubrics, getAgentvConfigDir, getAgentvHome, getOutputFilenames, getProject, getProjectsRegistryPath, getResultsRepoLocalPaths, getResultsRepoStatus, getSubagentsRoot, getTextContent, getTraceStateRoot, getWorkspacePath, getWorkspacePoolRoot, getWorkspacesRoot, groupTranscriptJsonLines, initializeBaseline, isAgentSkillsFormat, isContent, isContentArray, isGraderKind, isJsonObject, isJsonValue, isNonEmptyString, isTestMessage, isTestMessageRole, killAllTrackedChildren, listGitRuns, listTargetNames, loadConfig, loadEvalCaseById, loadEvalCases, loadEvalSuite, loadProjectRegistry, loadTestById, loadTestSuite, loadTests, loadTsConfig, loadTsEvalFile, materializeGitRun, mergeExecutionMetrics, negateScore, normalizeLineEndings, normalizeResultsConfig, parseAgentSkillsEvals, parseClaudeSession, parseCodexSession, parseCopilotEvents, parseEnvOutput, parseJsonFromText, parseJsonSafe, parseYamlValue, prepareResultsRepoBranch, pushResultsRepoBranch, readJsonFile, readTargetDefinitions, readTestSuiteMetadata, readTextFile, readTranscriptFile, readTranscriptJsonl, removeProject, resolveAndCreateProvider, resolveDelegatedTargetDefinition, resolveFileReference, resolveResultsRepoRunsDir, resolveResultsRepoUrl, resolveTargetDefinition, resolveWorkspaceTemplate, rubricEvaluationSchema, runBeforeSessionHook, runContainsAllAssertion, runContainsAnyAssertion, runContainsAssertion, runEndsWithAssertion, runEqualsAssertion, runEvalCase, runEvaluation, runIcontainsAllAssertion, runIcontainsAnyAssertion, runIcontainsAssertion, runIsJsonAssertion, runRegexAssertion, runStartsWithAssertion, saveProjectRegistry, scanRepoDeps, scoreRangeEvaluationSchema, scoreToVerdict, shouldEnableCache, shouldSkipCacheForTemperature, stageResultsArtifacts, subscribeToClaudeLogEntries, subscribeToCodexLogEntries, subscribeToCopilotCliLogEntries, subscribeToCopilotSdkLogEntries, subscribeToPiLogEntries, substituteVariables, syncProject, syncProjects, syncResultsRepo, toCamelCaseDeep, toSnakeCaseDeep, toTranscriptJsonLines, tokensPerTool, touchProject, trackChild, trackedChildCount, transpileEvalYaml, transpileEvalYamlFile, trimBaselineResult };