githubnext / githubnext/gh-aw-test
Suggested test: copilot/sandbox-agent-memory-limit
- Dominant language
- Shell
- Stars
- 8
- Forks
- 5
- PR merge metrics
- No merged PRs in 30d
Description
## Motivation
github/gh-aw#49448 — fixed `sandbox.agent.memory` frontmatter field being silently dropped (never reached `--memory-limit` in compiled AWF invocation). Also added compile-time format validation via `validateAgentMemoryLimit`.
## Proposed test
- **Workflow file**: `test-copilot-sandbox-agent-memory.md`
- **Trigger**: `workflow_dispatch`
- **Engine**: copilot
- **Safe output**: `create-issue`
- **Variant**: standard
## Minimal test prompt sketch
Set `sandbox.agent.memory: 2g` in frontmatter, then have the agent create an issue confirming it ran (a smoke test that the workflow compiles and runs successfully with the memory limit applied — similar in spirit to `test-copilot-restore-memory-custom-job.md` but for the sandbox memory limit rather than cache-memory).
## New fixtures or secrets needed
None.
## Notes
Not to be confused with `cache-memory`/`restore-memory` (already covered by `test-copilot-restore-memory-custom-job.md`). This targets the separate `sandbox.agent.memory` field that maps to `--memory-limit`. Primary value is confirming the lock file actually contains `--memory-limit 2g` and that the run doesn't fail (e.g. OOM/exit 137) — could be checked via a follow-up grep on the generated `.lock.yml` rather than requiring live verification.
> Generated by [🔍 Suggest New E2E Tests](https://github.com/githubnext/gh-aw-test/actions/runs/30693265278) · auto · 30.2 AIC · ⌖ 8.86 AIC · ⊞ 8.3K · [◷](https://github.com/search?q=repo%3Agithubnext%2Fgh-aw-test+is%3Aissue+%22gh-aw-workflow-call-id%3A+githubnext%2Fgh-aw-test%2Fsuggest-new-e2e-tests%22&type=issues)
Contributor guide
No contributing guide indexed for this repository
Research direction
Compare the proposed test-copilot-sandbox-agent-memory.md workflow with test-copilot-restore-memory-custom-job.md, focusing on the frontmatter, workflow_dispatch trigger, copilot engine, and create-issue output. Run the existing test workflow or its validation path, then confirm the generated .lock.yml contains --memory-limit 2g and the workflow completes successfully without new fixtures or secrets.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- github-actions
- Domain
- ci-cd, testing
- Issue type
- Feature
- Difficulty
- 2/5
- Estimated time
- 1-3 hours
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 74/100