Compare commits

...

2 Commits

Author SHA1 Message Date
shivammittal274
639b2e915e fix(eval): log agent errors in task progress output for CI visibility 2026-03-21 23:31:20 +05:30
shivammittal274
d050afa802 chore(eval): remove unused env vars from workflow (OPENROUTER, OPENAI) 2026-03-21 23:21:35 +05:30
2 changed files with 6 additions and 2 deletions

View File

@@ -49,8 +49,6 @@ jobs:
working-directory: packages/browseros-agent/apps/eval
env:
FIREWORKS_API_KEY: ${{ secrets.FIREWORKS_API_KEY }}
OPENROUTER_API_KEY: ${{ secrets.OPENROUTER_API_KEY }}
OPENAI_API_KEY: ${{ secrets.OPENAI_API_KEY }}
CLAUDE_CODE_OAUTH_TOKEN: ${{ secrets.CLAUDE_CODE_OAUTH_TOKEN }}
BROWSEROS_BINARY: /usr/bin/browseros
EVAL_CONFIG: ${{ github.event.inputs.config || 'configs/browseros-agent-weekly.json' }}

View File

@@ -242,6 +242,12 @@ function printTaskProgress(
if (result.status === 'failed') {
console.log(` ERROR: ${result.error.message}`)
} else if (isSuccessfulResult(result)) {
// Log agent errors (e.g., LLM API failures) even if task "completed"
if (result.agentResult.metadata.errors?.length) {
for (const err of result.agentResult.metadata.errors) {
console.log(` ERROR [${err.source}]: ${err.message}`)
}
}
for (const [name, gr] of Object.entries(result.graderResults)) {
const icon = gr.pass ? 'PASS' : 'FAIL'
console.log(` ${name}: ${icon}`)