Repository navigation
feat(e2b): support verified template compute profiles - #55
Conversation
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 2952be00df
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| case "$COMPUTE_PROFILE" in | ||
| 2c2g) ;; | ||
| 2c8g|4c8g) compute_args=(--compute-profile "$COMPUTE_PROFILE") ;; |
There was a problem hiding this comment.
When 2c2g is selected, this branch deliberately omits the flag to preserve compatibility with historical source parsers. That makes the operator’s selection depend on the checked-out builder’s implicit default; if a selected beta or later source changes that default, the subsequent check still accepts it because it compares E2B’s allocation with that same builder’s kwargs rather than with COMPUTE_PROFILE. Require the selected revision to support the interface and pass --compute-profile 2c2g explicitly instead of retaining this compatibility fallback.
AGENTS.md reference: AGENTS.md:L59-L61
Useful? React with 👍 / 👎.
The E2B template workflow can build a separate 2 vCPU / 8 GiB or 4 vCPU / 8 GiB template while keeping 2 vCPU / 2 GiB as its default. Fresh and resumed builds verify the exact immutable build's CPU and memory before upload and preserve the build receipt on validation failure. Default builds from historical source revisions still omit the new optional builder argument.
Validation: all 15 E2B Python unit tests pass on Python 3.14; seven workflow shell blocks and embedded Python compile; independent review found no remaining actionable findings. No live E2B build or deployment was performed. This does not change Core's deployed default resources; per-session selection is a separate change.