refactor(evals,web): Update consumers to handle PromptResult aborts

- Handle abort in evals/buffbench commit screening
- Use ABORT_ERROR_MESSAGE constant in web admin route
- Remove unused scripts/get-changelog.ts

Author: brandonkachen

evals/buffbench/pick-commits.ts

@@ -398,7 +398,7 @@ async function screenCommitsWithGpt5(
     const prompt = `${COMMIT_SCREENING_PROMPT}\n\nCommit to evaluate:\n\n${commitInfo}`
 
     try {
-      const response = await promptAiSdkStructured({
+      const result = await promptAiSdkStructured({
         messages: [userMessage(prompt)],
         schema: CommitSelectionSchema,
         model: models.openrouter_gpt5,
@@ -414,6 +414,14 @@ async function screenCommitsWithGpt5(
         signal: new AbortController().signal,
       })
 
+      // Handle aborted request
+      if (result.aborted) {
+        console.log(`Commit ${commit.sha.substring(0, 8)} screening aborted`)
+        return null
+      }
+
+      const response = result.value
+
       // Handle empty or invalid response
       if (
         !response ||

scripts/get-changelog.ts

@@ -164,7 +164,7 @@ Please create a changelog with:
 
 Start your response with a heading using ### (three hashes) and organize the content below it.`
 
-    const response = await promptAiSdk({
+    const result = await promptAiSdk({
       messages: [userMessage(prompt)],
       clientSessionId: generateCompactId(),
       fingerprintId: generateCompactId(),
@@ -180,6 +180,14 @@ Start your response with a heading using ### (three hashes) and organize the con
       signal: new AbortController().signal,
     })
 
+    // Handle aborted request
+    if (result.aborted) {
+      console.log(`⏹️  Changelog generation was aborted`)
+      return false
+    }
+
+    const response = result.value
+
     // Clean up the AI response
     console.log(`🧹 Cleaning up AI response...`)
     const cleanedText = response

web/src/app/api/admin/relabel-for-user/route.ts

@@ -15,6 +15,7 @@ import {
   models,
   TEST_USER_ID,
 } from '@codebuff/common/old-constants'
+import { unwrapPromptResult } from '@codebuff/common/util/error'
 import { userMessage } from '@codebuff/common/util/messages'
 import { generateCompactId } from '@codebuff/common/util/string'
 import { closeXml } from '@codebuff/common/util/xml'
@@ -206,11 +207,13 @@ async function relabelTraceWithModel(params: {
       system: payload.system as System,
     })
 
-    const output = await promptAiSdk({
-      ...promptContext,
-      model,
-      messages,
-    })
+    const output = unwrapPromptResult(
+      await promptAiSdk({
+        ...promptContext,
+        model,
+        messages,
+      }),
+    )
 
     const relabel: Relabel = {
       id: generateCompactId(),
@@ -350,12 +353,14 @@ async function relabelWithRelace(params: {
     filesWithPath.map((file) => `- ${file.path}`).join('\n'),
   ].join('\n\n')
 
-  const ranked = await promptAiSdk({
-    ...promptContext,
-    model: models.openrouter_claude_sonnet_4,
-    messages: [userMessage(prompt)],
-    includeCacheControl: false,
-  })
+  const ranked = unwrapPromptResult(
+    await promptAiSdk({
+      ...promptContext,
+      model: models.openrouter_claude_sonnet_4,
+      messages: [userMessage(prompt)],
+      includeCacheControl: false,
+    }),
+  )
 
   const rankedFiles =
     ranked
@@ -432,15 +437,17 @@ async function relabelWithClaudeWithFullFileContext(params: {
     system = systemCopy
   }
 
-  const output = await promptAiSdk({
-    ...promptContext,
-    model,
-    messages: messagesWithSystem({
-      messages: (tracePayload.messages || []) as Message[],
-      system,
+  const output = unwrapPromptResult(
+    await promptAiSdk({
+      ...promptContext,
+      model,
+      messages: messagesWithSystem({
+        messages: (tracePayload.messages || []) as Message[],
+        system,
+      }),
+      maxOutputTokens: 1000,
     }),
-    maxOutputTokens: 1000,
-  })
+  )
 
   const relabel: Relabel = {
     id: generateCompactId(),