fix(ui): surface a failed program launch where people actually are (#441) (v0.38.4) - #446
Merged
Merged
Conversation
) (v0.38.4) v0.38.0 made the API report programStarted:false with a reason. Nothing displayed it, so the dashboard still showed a healthy-looking agent and someone had to read the API response or the server log to find out. That is the half the reporter of #426 would actually have seen. - First-agent onboarding wizard: shown on the success screen. The most important of the three — it is someone's first agent, and #426 was reported by a user in exactly that position whose first message was executed by bash. The heading changes from "ready to help you build amazing things" to "created, but its AI program did not start". - Agent creation wizard: same warning on its success screen. The agent genuinely exists, so a warning rather than a failure. - Wake: a 15s warning toast. The wake path only checked !response.ok, so HTTP 200 with programStarted:false was treated as complete success and said nothing. All three name the consequence, not just the fault: the terminal is at a shell prompt, so anything sent now would be run as a shell command instead of read. Verified end to end by creating an agent configured with aider, which is not installed here: { "success": true, "programStarted": false, "programError": "\"aider\" did not start: the program is not on PATH ...", "shellSaid": "zsh: command not found: aider" } Control with an installed program: programStarted:true, no warning. Closes the case that began with a fresh install and ends with a configured-but-missing framework: choosing aider, codex, cursor or a wrapper you have not installed used to be indistinguishable from a working agent at every layer — service, API and UI. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
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
1 participant
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.
Closes #441.
v0.38.0 made the API report
programStarted: falsewith a reason. Nothing displayed it. The dashboard still showed a healthy-looking agent, and finding out meant reading the API response or the server log — which is the half the reporter of #426 would actually have seen.Three surfaces
The first-agent onboarding wizard — the most important of the three. It is someone's very first agent, and #426 was reported by a user in exactly that position whose first message was executed by bash. The heading changes from "ready to help you build amazing things" to "created, but its AI program did not start".
The agent creation wizard — same warning on its success screen. The agent genuinely exists, so this is a warning, not a failure, and the wizard stops implying everything is fine.
Waking an agent — a 15-second warning toast. This path only checked
!response.ok, so a wake returning HTTP 200 withprogramStarted: falsewas treated as complete success and said nothing at all.All three name the consequence, not just the fault:
Verified end to end
Creating an agent configured with
aider, which is not installed on this machine — the exact scenario suspected when this was raised ("maybe the user tried to create an agent using a framework he did not have installed?"):{ "success": true, "programStarted": false, "programError": "\"aider\" did not start: the program is not on PATH inside the tmux session. tmux starts a non-login shell, so a PATH exported from ~/.bash_profile or ~/.profile is not present — move it to ~/.bashrc (or ~/.zshrc), or check the program is installed at all.", "shellSaid": "zsh: command not found: aider" }Control, with a program that is installed:
programStarted: true, no warning. Both test agents cleaned up.What this closes
The case ran from a fresh install (#426) to a configured-but-missing framework. Choosing
aider,codex,cursoror asandbox-execwrapper you have not installed used to be indistinguishable from a working agent at every layer — service, API, and UI. All three now say so.1257 tests passing.
🤖 Generated with Claude Code