[buffbench] add more logging to eval errors

Author: charleslien

evals/git-evals/run-git-evals.ts

@@ -4,6 +4,7 @@ import path from 'path'
 
 import { disableLiveUserInputCheck } from '@codebuff/backend/live-user-inputs'
 import { promptAiSdkStructured } from '@codebuff/backend/llm-apis/vercel-ai-sdk/ai-sdk'
+import { errorToObject } from '@codebuff/common/util/object'
 import { withTimeout } from '@codebuff/common/util/promise'
 import { generateCompactId } from '@codebuff/common/util/string'
 import { cloneDeep } from 'lodash'
@@ -247,7 +248,7 @@ Explain your reasoning in detail.`,
     return {
       ...evalRun,
       judging_results: {
-        analysis: 'Judging failed due to error',
+        analysis: `Judging failed due to error:\n${judgingError instanceof Error ? errorToObject(judgingError) : JSON.stringify(judgingError)}`,
         strengths: [],
         weaknesses: ['Judging process encountered an error'],
         metrics: {