Problem

When apostCreateCommand fails in adevcontainer.json, the current experience is not ideal for users. Even though we receive acontainerId we're not handling this gracefully.

Current Behavior

When there's a script error,devcontainer up outputs:

{"outcome":"error","message":"Command failed: /bin/sh -c exit 1","description":"postCreateCommand from devcontainer.json failed.","containerId":"86fb694d78cadde3e5f7e1af75b3675ba623282defe763e79d77d620355502d1"}

Currently we depend onoutcome, but we still get thecontainerId which is all that really matters for us.

Desired Behavior

• Log the error but still inject if we get a container ID
• Improve error surfacing in UI (currently limited to showing just one error, could show multiple)

Context

One of our customers ran into this issue and was unable to connect to their devcontainer because we didn't inject the agent even though the devcontainer was running.

Related Discussion

https://coder.slack.com/archives/C06UK9XP3F0/p1763486241456639