Skip to content

Updated security campaign worker workflows to discover and address all security alerts (not just critical/high severity) - #12005

Merged
mnkiefer merged 6 commits into
mainfrom
copilot/fix-security-alerts-discrepancy
Jan 27, 2026
Merged

mnkiefer merged 6 commits into
mainfrom
copilot/fix-security-alerts-discrepancy

Conversation

Copilot AI commented Jan 27, 2026 •

Copy link
Copy Markdown
Contributor

Update Workers to Discover All Security Alerts ✅

Problem Solved

Workers were only targeting critical and high severity alerts, missing alerts with other severities. Out of 22 total security alerts, workers could only address the critical/high subset due to severity filtering.

Solution Implemented

  • Updated code-scanning-fixer.md to discover all severity levels
    • Changed description from "critical and high severity" to "all severity levels"
    • Updated Step 2 to list ALL open alerts (removed severity filter)
    • Updated Step 3 to prioritize by severity but not exclude any
    • Updated exit messages to reflect all severities
    • Updated security guidelines and error handling sections
  • Reviewed security-fix-pr.md (already supports all severities - no changes needed)
  • Recompiled workflow lock files (139/140 compiled successfully)
  • Formatted code
  • JavaScript linting passed
  • Code review completed - no issues found
  • Security scan (CodeQL) - no alerts

Key Changes

Before:

List critical and high severity alerts:
- Call with severity: "critical"
- Call with severity: "high"
- Combine results
- Exit if no critical/high found

After:

List all open alerts:
- Call without severity filter
- Sort by severity (critical > high > medium > low > warning > note > error)
- Process highest severity unfixed alert
- Exit if no alerts found

Impact

The code-scanning-fixer worker will now:

  1. ✅ Discover and address all 22 security alerts (not just critical/high)
  2. ✅ Still prioritize by severity (critical first, then high, then medium, etc.)
  3. ✅ Create labeled PRs for all alert types (z_campaign_security-alert-burndown)
  4. ✅ Enable discovery system to find all 22 alerts as workers create PRs over time

Architecture

This maintains the correct label-based discovery architecture:

  • Workers create PRs for alerts and label them
  • Discovery finds labeled PRs automatically
  • No hard-coded alert types in discovery system
  • Campaign-agnostic, flexible design

Validation

  • ✅ Workflows recompiled successfully
  • ✅ JavaScript formatting validated
  • ✅ Campaign spec validation passed
  • ✅ Code review completed - no issues
  • ✅ Security scan - no alerts

Security Summary

No security vulnerabilities introduced. Changes are documentation/instruction updates to the worker workflow markdown files, enabling broader alert coverage while maintaining security best practices and severity-based prioritization.

Original prompt

Based on https://github.com/githubnext/gh-aw/security there should be 22 security alerts, but the security alert burndown campaign only discovered 5 items (1 Epic issue, 4 security fix PRs), see:
https://github.com/githubnext/gh-aw/actions/runs/21389242618/job/61572645950


💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

Loading
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants