Repository navigation
Conversation
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
|
Review the following changes in direct dependencies. Learn more about Socket for GitHub.
|
msullivan
force-pushed
the
remote-agent-2
branch
from
November 13, 2026 21:06
6fedd51 to
34d8328
Compare
msullivan
force-pushed
the
remote-agent-2
branch
from
November 13, 2026 23:30
34d8328 to
6eb6470
Compare
msullivan
force-pushed
the
remote-agent-2
branch
from
November 14, 2026 01:01
6eb6470 to
75fcdf4
Compare
msullivan
force-pushed
the
remote-agent-2
branch
from
November 14, 2026 01:33
75fcdf4 to
7959f6e
Compare
msullivan
force-pushed
the
remote-agent-2
branch
from
November 14, 2026 01:39
7959f6e to
d8a47a5
Compare
msullivan
force-pushed
the
remote-agent-2
branch
from
November 14, 2026 01:44
d8a47a5 to
9ad8765
Compare
msullivan
marked this pull request as ready for review
November 16, 2026 23:09
msullivan
force-pushed
the
remote-agent-2
branch
from
November 17, 2026 00:18
494bf8e to
2bc286f
Compare
I am not sure what library this code actually belongs in, but it is glue for allowing tools/loops to be automatically run on Vercel Sandbox. None of this code is really ai specific. There are two main components: * `ai.contrib.sandbox.runner`: generic runner and client code for running python functions in another process (possibly a remote one on a sandbox). This is sandbox agnostic and just depends on being handed anyio streams for talking to a runner process. * `ai.contrib.sandbox.vercel`: which provides glue for syncing the calling project over to a provided vercel sandbox and hooks up streams to pass to the runner client. (The stream part is actually quite annoying because the sandbox API mangles binary, so we need to base64 everything.) The runner supports running both sync and async functions as well as generators. Results may return out of order. By default generators need to be advanced one element at a time but a "free running" mode can be requested where generators will generate elements without waiting for a request. To avoid either blowing up memory or blocking everything on one slow generator consumer, we have per-generator backpressure.
msullivan
force-pushed
the
remote-agent-2
branch
from
November 17, 2026 01:16
2bc286f to
af55f3b
Compare
This branch was successfully deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
I am not sure what library this code actually belongs in, but it is glue for allowing tools/loops to be automatically run on Vercel Sandbox. None of this code is really ai specific.
There are two main components:
ai.contrib.sandbox.worker: generic worker and client code for running python functions in another process (possibly a remote one on a sandbox). This is sandbox agnostic and just depends on being handed anyio streams for talking to a worker process.ai.contrib.sandbox.vercel: which provides glue for syncing the calling project over to a provided vercel sandbox and hooks up streams to pass to the worker client. (The stream part is actually quite annoying because the sandbox API mangles binary, so we need to base64 everything.)The runner supports running both sync and async functions as well as generators. Results may return out of order. By default generators need to be advanced one element at a time but a "free running" mode can be requested where generators will generate elements without waiting for a request. To avoid either blowing up memory or blocking everything on one slow generator consumer, we have per-generator backpressure.