E2E status report — 2026-06-22 (gh-aw matrix) #3129

You are triaging failures from the gh-aw-test E2E suite.
Run: https://github.com/githubnext/gh-aw-test/actions/runs/27957605974
Repository under test: github/gh-aw (the gh-aw CLI/compiler).
Test harness repository: githubnext/gh-aw-test (this repo; runner is e2e.sh).

Goal: for EACH failed test listed in this status report, access the GitHub
Actions logs for the run above (and the per-entry artifacts e2e-<label>-samples-<bool>,
which contain e2e-test-*.log, e2e-output.log and fails.txt), determine the root
cause, and categorize the failure as exactly one of:

  1. TRANSIENT — flaky/infra/network/rate-limit/timing; not a real defect.
     Action: note it and recommend a re-run (./e2e.sh rerun).
  2. TEST-FRAMEWORK BUG — a defect in this repo's harness (e2e.sh), a workflow
     source file (.github/workflows/test-*.md), a sample, or CI config.
     Action: propose a concrete fix (file + change) in githubnext/gh-aw-test.
  3. GH-AW BUG — a defect in github/gh-aw itself (compiler output, runtime
     engine behaviour, safe-output handling, etc.).
     Action: open an issue in github/gh-aw with a minimal repro, the failing
     test name, the gh-aw ref/mode/samples combination, and links to the
     relevant log lines. Check for an existing open issue first and link it
     instead of filing a duplicate.

Steps:
  - Use 'gh run view <run-id> --log' and 'gh run download <run-id>' to fetch logs/artifacts.
  - Read AGENTS.md in githubnext/gh-aw-test for harness conventions before proposing fixes.
  - Group failures by suspected root cause; the same gh-aw bug may explain several.
  - Produce a table: test | category | root cause | recommended action | issue/PR link.
  - Only open github/gh-aw issues for category 3, and only after confirming no duplicate exists.