@wix/evalforge-types 0.39.0 → 0.40.0

package/build/index.mjs

@@ -1363,8 +1363,6 @@ var LlmJudgeAssertionSchema = z21.object({
   type: z21.literal("llm_judge"),
   /** Prompt template; placeholders: {{output}}, {{cwd}}, {{changedFiles}}, {{trace}} */
   prompt: z21.string(),
-  /** Optional system prompt for the judge (default asks for JSON with score) */
-  systemPrompt: z21.string().optional(),
   /** Minimum score to pass (0-100, default 70) */
   minScore: z21.number().int().min(0).max(100).optional(),
   /** Model for the judge (e.g. claude-3-5-haiku) */
@@ -1497,8 +1495,6 @@ var LlmJudgeConfigSchema = z23.object({
    * - Custom parameters defined in the parameters array
    */
   prompt: z23.string().min(1),
-  /** Optional system prompt for the judge */
-  systemPrompt: z23.string().optional(),
   /** Minimum score to pass (0-100, default 70) */
   minScore: z23.number().int().min(0).max(100).optional(),
   /** Model for the judge (e.g. claude-3-5-haiku-20241022) */
@@ -2178,19 +2174,6 @@ var SYSTEM_ASSERTIONS = {
         required: true,
         defaultValue: "Verify the output meets the acceptance criteria."
       },
-      {
-        name: "systemPrompt",
-        label: "System Prompt (optional)",
-        type: "string",
-        required: false,
-        defaultValue: `You are judging a scenario run. Use these values:
-- {{output}}: the agent's final output
-- {{cwd}}: working directory
-- {{changedFiles}}: list of files changed (or "No files were changed")
-- {{trace}}: step-by-step trace (tool calls, completions) to check e.g. which tools were called and how many times
-
-Judge how well the output meets the acceptance criteria stated in the user prompt.`
-      },
       {
         name: "minScore",
         label: "Minimum Score (0-100)",