P1 Systemic: Multiple workflows failing with "engine terminated unexpectedly" (exit code 1) — Apr 8

[github-actions]

Overview

20 workflow failure issues were created on Apr 8, 2026 — a significant spike. At least 13 workflows share a common failure pattern: the engine terminates with exit code 1 after the firewall containers stop cleanly. This spans both copilot and claude engines.

Failure Pattern

The common log signature is:

Container awf-squid  Removing
Container awf-squid  Removed
[SUCCESS] Containers stopped successfully
[INFO] Agent session state preserved at: /tmp/awf-agent-session-state-...
[INFO] API proxy logs available at: /tmp/gh-aw/sandbox/firewall/logs/api-proxy-logs
[WARN] Command completed with exit code: 1
Process exiting with code: 1

The container/firewall infrastructure teardown succeeds, but the agent process exits with code 1, causing GitHub Actions to report the workflow as failed.

Affected Workflows (Copilot engine, exit code 1)

Issue	Workflow	Time (UTC)
#25287	Dead Code Removal Agent	12:04
#25282	Daily Copilot Token Usage Audit	11:55
#25277	Daily Testify Uber Super Expert	11:39
#25268	Copilot PR Conversation NLP Analysis	10:43
#25267	Daily Syntax Error Quality Check	10:41
#25262	Daily MCP Tool Concurrency Analysis	09:36
#25261	Dev	09:05
#25260	Architecture Diagram Generator	09:03
#25249	Daily CLI Performance Agent	05:45
#25243	Copilot PR Prompt Pattern Analysis	05:02
#25236	Agent Performance Analyzer	04:37
#25215	Auto-Triage Issues	01:02

Also affected (Claude engine):

• [aw] CI Cleaner failed #25257 CI Cleaner (06:54) — same exit code 1 pattern

Timeline & Possible Causes

Failures span 01:02–12:04 UTC, beginning before any code changes landed today:

Time	Event
01:02	First failures (Auto-Triage, Smoke tests) — before any commits
04:37	Agent Performance Analyzer (Copilot) — still before commits
04:46	feat: repo-level config via aw.json merged
05:05	chore: bump firewall v0.25.16 + full recompile
05:57	feat: add pre-steps to agent job merged
06:08	chore: Copilot CLI 1.0.20→1.0.21 merged
06:54	CI Cleaner (Claude engine) fails
09:03–12:04	Wave of Copilot engine failures

Since failures started before any commits, this may be an infrastructure or GitHub Actions Copilot runner issue. The post-05:05 failures may or may not be related to the code changes.

Possible Causes to Investigate

1. GitHub Copilot runner infrastructure: Agent process failing silently (exit code 1) without useful error output — check if there's a known outage or degradation
2. Firewall v0.25.16: The recompile happened at 05:05 — check if new firewall version blocks something the agent requires
3. Copilot CLI 1.0.21: New version (06:08) — check release notes for breaking changes that could cause exit code 1
4. pre-steps feature: New same-job token minting (05:57) — may affect auth flow for some workflows

Investigation Steps

1. Download agent-stdio.log from a failed run (e.g., §24134231123) to see what the Copilot agent logged before exit
2. Check GitHub Status page for Copilot runner issues
3. Compare a failing workflow's lock file (pre vs post firewall bump) to see if behavior changed
4. Check if Copilot CLI 1.0.21 has known exit code 1 issues

Priority

P1 — Affects 13+ workflows including core agentic workflows (Dev, Agent Performance Analyzer, Auto-Triage Issues).

Detected by Workflow Health Manager meta-orchestrator (§24134411505).

Generated by Workflow Health Manager - Meta-Orchestrator · ● 1.6M · ◷

• expires on Apr 9, 2026, 12:15 PM UTC

[github-actions]

🍪 Issue Monster has assigned this to Copilot!

I've identified this issue as a good candidate for automated resolution and assigned it to the Copilot coding agent.

The Copilot coding agent will analyze the issue and create a pull request with the fix.

Om nom nom! 🍪

🍪 Om nom nom by Issue Monster · ● 169.3K · ◷