Skip to content

Auto Fix Failed Tests #2792

Auto Fix Failed Tests

Auto Fix Failed Tests #2792

name: Auto Fix Failed Tests
on:
workflow_run:
workflows: ["PR Checks", "CLI PR Checks"]
types:
- completed
permissions:
contents: write
pull-requests: write
issues: write
actions: read
jobs:
fix-failed-tests:
# Only run if the workflow failed
# DISABLED: Remove 'false && ' on the next line to enable auto-fixing
if: false && github.event.workflow_run.conclusion == 'failure'
runs-on: ubuntu-latest
steps:
- name: Get workflow run details
id: workflow-details
uses: actions/github-script@v8
with:
script: |
const workflowRun = context.payload.workflow_run;
const { data: jobs } = await github.rest.actions.listJobsForWorkflowRun({
owner: context.repo.owner,
repo: context.repo.repo,
run_id: workflowRun.id
});
// Find all failed jobs since we're only monitoring specific test workflows
const failedJobs = jobs.jobs.filter(job => job.conclusion === 'failure');
if (failedJobs.length === 0) {
core.info('No failed jobs found');
return null;
}
core.setOutput('has_failed_tests', 'true');
core.setOutput('workflow_name', workflowRun.name);
core.setOutput('workflow_run_id', workflowRun.id);
core.setOutput('head_branch', workflowRun.head_branch);
core.setOutput('head_sha', workflowRun.head_sha);
core.setOutput('failed_jobs', JSON.stringify(failedJobs.map(j => j.name)));
return failedJobs;
- name: Get job logs for failed tests
if: steps.workflow-details.outputs.has_failed_tests == 'true'
id: get-logs
uses: actions/github-script@v8
with:
script: |
const workflowRunId = ${{ github.event.workflow_run.id }};
const { data: jobs } = await github.rest.actions.listJobsForWorkflowRun({
owner: context.repo.owner,
repo: context.repo.repo,
run_id: workflowRunId
});
let errorLogs = '';
for (const job of jobs.jobs) {
if (job.conclusion === 'failure') {
try {
const { data: logData } = await github.rest.actions.downloadJobLogsForWorkflowRun({
owner: context.repo.owner,
repo: context.repo.repo,
job_id: job.id
});
errorLogs += `\n\n=== Job: ${job.name} ===\n`;
errorLogs += logData;
} catch (error) {
core.warning(`Could not fetch logs for job ${job.name}: ${error.message}`);
}
}
}
// Store logs in environment file for next step
const fs = require('fs');
fs.writeFileSync('/tmp/test_failure_logs.txt', errorLogs);
core.setOutput('has_logs', errorLogs.length > 0 ? 'true' : 'false');
- name: Checkout repository
if: steps.workflow-details.outputs.has_failed_tests == 'true'
uses: actions/checkout@v5
with:
token: ${{ secrets.GITHUB_TOKEN }}
fetch-depth: 0
ref: ${{ steps.workflow-details.outputs.head_sha }}
- name: Setup Node.js
if: steps.workflow-details.outputs.has_failed_tests == 'true'
uses: actions/setup-node@v5
with:
node-version: "20"
- name: Install Continue CLI globally
if: steps.workflow-details.outputs.has_failed_tests == 'true'
run: npm i -g @continuedev/cli
- name: Start remote session to fix failed tests
if: steps.workflow-details.outputs.has_failed_tests == 'true'
id: remote-session
env:
CONTINUE_API_KEY: ${{ secrets.CONTINUE_API_KEY }}
run: |
# Create a detailed prompt for fixing the failed tests
cat > /tmp/fix_tests_prompt.txt << 'PROMPT_EOF'
🔧 **Auto Test Fix Request**
The following tests failed in workflow "${{ steps.workflow-details.outputs.workflow_name }}" (Run ID: ${{ steps.workflow-details.outputs.workflow_run_id }}):
**Failed Jobs:** ${{ steps.workflow-details.outputs.failed_jobs }}
**Branch:** ${{ steps.workflow-details.outputs.head_branch }}
**Commit:** ${{ steps.workflow-details.outputs.head_sha }}
**Your Task:**
1. Analyze the test failure logs and error messages
2. Identify the root cause of the test failures
3. Fix the failing tests by updating the test code or the underlying implementation
4. Ensure all tests pass after your changes
5. Commit your fixes with a descriptive message
**Test Failure Context:**
Please examine the repository structure, run the failing tests locally to understand the errors, and implement appropriate fixes.
Focus on:
- Understanding what the tests are trying to validate
- Identifying why they're failing (code changes, environment issues, test logic errors)
- Making minimal, targeted fixes that address the root cause
- Ensuring the fixes don't break other functionality
Please start by examining the failing tests and their error messages, then proceed with the necessary fixes.
PROMPT_EOF
echo "Starting Continue CLI remote session for test fixes..."
echo "Prompt content:"
cat /tmp/fix_tests_prompt.txt
# Start remote session and capture JSON output
SESSION_OUTPUT=$(cat /tmp/fix_tests_prompt.txt | cn remote -s --branch ${{ steps.workflow-details.outputs.head_branch }})
echo "Raw session output: $SESSION_OUTPUT"
# Extract URL from JSON output
SESSION_URL=$(echo "$SESSION_OUTPUT" | jq -r '.url // empty')
if [ -z "$SESSION_URL" ] || [ "$SESSION_URL" = "null" ]; then
echo "Failed to extract session URL from output: $SESSION_OUTPUT"
exit 1
fi
echo "session_url=$SESSION_URL" >> $GITHUB_OUTPUT
echo "✅ Started remote session: $SESSION_URL"
- name: Log session details
if: steps.workflow-details.outputs.has_failed_tests == 'true'
run: |
echo "✅ Successfully started auto-fix session for failed tests"
echo "Workflow: ${{ steps.workflow-details.outputs.workflow_name }}"
echo "Run ID: ${{ steps.workflow-details.outputs.workflow_run_id }}"
echo "Branch: ${{ steps.workflow-details.outputs.head_branch }}"
echo "Session URL: ${{ steps.remote-session.outputs.session_url }}"
echo "Failed jobs: ${{ steps.workflow-details.outputs.failed_jobs }}"