@agentv/core 0.6.1 → 0.7.0

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.
Files changed (11) hide show
  1. package/dist/{chunk-OW3SHBIJ.js → chunk-L7I5UTJU.js} +1 -1
  2. package/dist/{chunk-OW3SHBIJ.js.map → chunk-L7I5UTJU.js.map} +1 -1
  3. package/dist/evaluation/validation/index.cjs.map +1 -1
  4. package/dist/evaluation/validation/index.js +1 -1
  5. package/dist/index.cjs +221 -242
  6. package/dist/index.cjs.map +1 -1
  7. package/dist/index.d.cts +11 -13
  8. package/dist/index.d.ts +11 -13
  9. package/dist/index.js +222 -243
  10. package/dist/index.js.map +1 -1
  11. package/package.json +1 -1
package/dist/index.d.cts CHANGED
@@ -99,17 +99,18 @@ type EvaluatorConfig = CodeEvaluatorConfig | LlmJudgeEvaluatorConfig;
99
99
  */
100
100
  interface EvalCase {
101
101
  readonly id: string;
102
- readonly dataset: string;
102
+ readonly dataset?: string;
103
103
  readonly conversation_id?: string;
104
- readonly task: string;
105
- readonly user_segments: readonly JsonObject[];
104
+ readonly question: string;
105
+ readonly input_segments: readonly JsonObject[];
106
+ readonly output_segments: readonly JsonObject[];
106
107
  readonly system_message?: string;
107
- readonly expected_assistant_raw: string;
108
+ readonly reference_answer: string;
108
109
  readonly guideline_paths: readonly string[];
109
110
  readonly guideline_patterns?: readonly string[];
110
111
  readonly file_paths: readonly string[];
111
112
  readonly code_snippets: readonly string[];
112
- readonly outcome: string;
113
+ readonly expected_outcome: string;
113
114
  readonly evaluator?: EvaluatorKind;
114
115
  readonly evaluators?: readonly EvaluatorConfig[];
115
116
  }
@@ -118,12 +119,12 @@ interface EvalCase {
118
119
  */
119
120
  interface EvaluationResult {
120
121
  readonly eval_id: string;
121
- readonly dataset: string;
122
+ readonly dataset?: string;
122
123
  readonly conversation_id?: string;
123
124
  readonly score: number;
124
125
  readonly hits: readonly string[];
125
126
  readonly misses: readonly string[];
126
- readonly model_answer: string;
127
+ readonly candidate_answer: string;
127
128
  readonly expected_aspect_count: number;
128
129
  readonly target: string;
129
130
  readonly timestamp: string;
@@ -167,7 +168,7 @@ declare function loadEvalCases(evalFilePath: string, repoRoot: URL | string, opt
167
168
  * Build prompt inputs by consolidating user request context and guideline content.
168
169
  */
169
170
  declare function buildPromptInputs(testCase: EvalCase): Promise<{
170
- request: string;
171
+ question: string;
171
172
  guidelines: string;
172
173
  systemMessage?: string;
173
174
  }>;
@@ -204,7 +205,7 @@ declare function resolveFileReference(rawValue: string, searchRoots: readonly st
204
205
  type ChatPrompt = AxChatRequest["chatPrompt"];
205
206
  type ProviderKind = "azure" | "anthropic" | "gemini" | "codex" | "cli" | "mock" | "vscode" | "vscode-insiders";
206
207
  interface ProviderRequest {
207
- readonly prompt: string;
208
+ readonly question: string;
208
209
  readonly guidelines?: string;
209
210
  readonly guideline_patterns?: readonly string[];
210
211
  readonly chatPrompt?: ChatPrompt;
@@ -404,7 +405,7 @@ interface EvaluationContext {
404
405
  readonly provider: Provider;
405
406
  readonly attempt: number;
406
407
  readonly promptInputs: {
407
- readonly request: string;
408
+ readonly question: string;
408
409
  readonly guidelines: string;
409
410
  readonly systemMessage?: string;
410
411
  };
@@ -442,10 +443,7 @@ declare class LlmJudgeEvaluator implements Evaluator {
442
443
  private readonly customPrompt?;
443
444
  constructor(options: LlmJudgeEvaluatorOptions);
444
445
  evaluate(context: EvaluationContext): Promise<EvaluationScore>;
445
- private evaluateWithAx;
446
446
  private evaluateWithPrompt;
447
- private buildJudgeForwardOptions;
448
- private buildJudgeModelConfig;
449
447
  }
450
448
  interface CodeEvaluatorOptions {
451
449
  readonly script: string;
package/dist/index.d.ts CHANGED
@@ -99,17 +99,18 @@ type EvaluatorConfig = CodeEvaluatorConfig | LlmJudgeEvaluatorConfig;
99
99
  */
100
100
  interface EvalCase {
101
101
  readonly id: string;
102
- readonly dataset: string;
102
+ readonly dataset?: string;
103
103
  readonly conversation_id?: string;
104
- readonly task: string;
105
- readonly user_segments: readonly JsonObject[];
104
+ readonly question: string;
105
+ readonly input_segments: readonly JsonObject[];
106
+ readonly output_segments: readonly JsonObject[];
106
107
  readonly system_message?: string;
107
- readonly expected_assistant_raw: string;
108
+ readonly reference_answer: string;
108
109
  readonly guideline_paths: readonly string[];
109
110
  readonly guideline_patterns?: readonly string[];
110
111
  readonly file_paths: readonly string[];
111
112
  readonly code_snippets: readonly string[];
112
- readonly outcome: string;
113
+ readonly expected_outcome: string;
113
114
  readonly evaluator?: EvaluatorKind;
114
115
  readonly evaluators?: readonly EvaluatorConfig[];
115
116
  }
@@ -118,12 +119,12 @@ interface EvalCase {
118
119
  */
119
120
  interface EvaluationResult {
120
121
  readonly eval_id: string;
121
- readonly dataset: string;
122
+ readonly dataset?: string;
122
123
  readonly conversation_id?: string;
123
124
  readonly score: number;
124
125
  readonly hits: readonly string[];
125
126
  readonly misses: readonly string[];
126
- readonly model_answer: string;
127
+ readonly candidate_answer: string;
127
128
  readonly expected_aspect_count: number;
128
129
  readonly target: string;
129
130
  readonly timestamp: string;
@@ -167,7 +168,7 @@ declare function loadEvalCases(evalFilePath: string, repoRoot: URL | string, opt
167
168
  * Build prompt inputs by consolidating user request context and guideline content.
168
169
  */
169
170
  declare function buildPromptInputs(testCase: EvalCase): Promise<{
170
- request: string;
171
+ question: string;
171
172
  guidelines: string;
172
173
  systemMessage?: string;
173
174
  }>;
@@ -204,7 +205,7 @@ declare function resolveFileReference(rawValue: string, searchRoots: readonly st
204
205
  type ChatPrompt = AxChatRequest["chatPrompt"];
205
206
  type ProviderKind = "azure" | "anthropic" | "gemini" | "codex" | "cli" | "mock" | "vscode" | "vscode-insiders";
206
207
  interface ProviderRequest {
207
- readonly prompt: string;
208
+ readonly question: string;
208
209
  readonly guidelines?: string;
209
210
  readonly guideline_patterns?: readonly string[];
210
211
  readonly chatPrompt?: ChatPrompt;
@@ -404,7 +405,7 @@ interface EvaluationContext {
404
405
  readonly provider: Provider;
405
406
  readonly attempt: number;
406
407
  readonly promptInputs: {
407
- readonly request: string;
408
+ readonly question: string;
408
409
  readonly guidelines: string;
409
410
  readonly systemMessage?: string;
410
411
  };
@@ -442,10 +443,7 @@ declare class LlmJudgeEvaluator implements Evaluator {
442
443
  private readonly customPrompt?;
443
444
  constructor(options: LlmJudgeEvaluatorOptions);
444
445
  evaluate(context: EvaluationContext): Promise<EvaluationScore>;
445
- private evaluateWithAx;
446
446
  private evaluateWithPrompt;
447
- private buildJudgeForwardOptions;
448
- private buildJudgeModelConfig;
449
447
  }
450
448
  interface CodeEvaluatorOptions {
451
449
  readonly script: string;