Repository navigation
Conversation
…a bundle is
First live runs of App Studio on the desktop development build, with the
system agent (an OpenAI model) driving the tools from a plain brief:
- `studio.image`: the model sent a job of its own making ("1", "2", "new",
a name) beside every prompt, six times over two turns, even after being
told not to; the refusal taught it nothing and no image was ever asked
for. A call with a prompt now starts a generation whatever else it
carries, only a call without one polls; every property has a
description and the tool text says the tool assigns the job id.
- `studio.bundle_check`: the model passed `main.splash` itself, then a
folder without a manifest, and got "manifest.json: No such file…". The
error and the tool text now say a bundle is a folder in the workspace
holding manifest.json and main.splash.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This branch has not been 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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Two tool-contract fixes from the first live App Studio runs on the desktop development build, where the system agent (an OpenAI model) drove the studio tools from a plain brief (observed, not scripted):
studio.imagetook no image. The model sent ajobof its own making ("1","2","new", then a name) beside every prompt, six times over two turns, even after a user message told it to send only the prompt. The tool refused each call ("give prompt to start an image, or job alone to poll one") and no generation was ever requested. Now a call with apromptstarts a generation whatever else it carries, and only a call without a prompt polls; every input property carries a description, and the tool text says the tool assigns the job id and that a prompt always starts a new generation.studio.bundle_checkdid not say what a bundle is. The model passedmain.splashitself, then a folder without a manifest, and got "manifest.json: No such file or directory". The manifest error and the tool text now say a bundle is a folder in the workspace holdingmanifest.jsonandmain.splash, passed asbundle_path.No behaviour changes for a correct caller. Docs are unchanged: the walkthrough's row already describes polling with the job value alone.
Still open from those runs, not in this PR: the system agent has no Splash language reference or bundle contract to read, so it wrote
main.splashin an invented syntax; that needs a guide the studio hands the agent (a tool or a skill), which is a design decision.Tests
--features mobile-apps,dev-mode -p octosense-shell studio: 47 passed (the happy-path test now starts with a prompt plus a strayjob, and the schema test accepts that shape).--features app-hub studio: 47 passed.🤖 Generated with Claude Code