feat: export eval artifacts as json#186
Conversation
Signed-off-by: Jeremy lewi <jeremy@lewi.us>
Signed-off-by: Jeremy lewi <jeremy@lewi.us>
Signed-off-by: Jeremy lewi <jeremy@lewi.us>
Signed-off-by: Jeremy lewi <jeremy@lewi.us>
Signed-off-by: Jeremy lewi <jeremy@lewi.us>
Signed-off-by: Jeremy lewi <jeremy@lewi.us>
![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: 82479c38d8
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".
| target.handle && | ||
| typeof target.handle.uri === "string" && | ||
| target.handle.uri.trim() |
There was a problem hiding this comment.
This resolver now falls back to the current notebook whenever target is present but has an empty uri/handle.uri, which changes the previous behavior (returning null) into implicit mutation of whatever notebook is currently open. If a tool call or generated script passes { uri: "" } (or an empty handle URI), runme.* operations can act on the wrong notebook instead of failing fast, which can silently corrupt eval results and user state.
Useful? React with 👍 / 👎.
| runOrThrow(`agent-browser open ${FRONTEND_URL}`, 30000); | ||
| run("agent-browser wait 2500", 5000); | ||
| evaluateJson<boolean>( | ||
| `JSON.stringify(await window.__runmeEval.waitUntilReady())`, |
There was a problem hiding this comment.
The runner invokes window.__runmeEval.waitUntilReady() immediately after a fixed sleep, but __runmeEval is installed from a React useEffect and is not guaranteed to exist by then. On slower page loads this throws before readiness polling even starts, causing flaky hard failures in the eval pipeline; the script should first poll for window.__runmeEval itself, then call waitUntilReady.
Useful? React with 👍 / 👎.
1 participant
Summary
Validation