web-infra-dev / web-infra-dev/midscene
[Feature]: Allow Studio to pass advanced Agent and report settings to platform runtimes
Nobody has claimed this yet.
- Dominant language
- TypeScript
- Stars
- 14.9k
- Forks
- 1.2k
- Avg merge
- 1d 14h
- Merged PRs (30d)
- 96
Description
What problem does this feature solve?
Midscene Studio currently provides model environment configuration, but most
runtime options supported by AgentOpt are not configurable from Studio.
Studio wraps the Web, Android, iOS, HarmonyOS, and Computer playground runtimes.
When creating or recreating an Agent, these runtimes mostly use their default
options. As a result, users cannot use existing Agent capabilities such as:
- replanning cycle limit
- delay after each action
- screenshot shrink factor
- AI action context
- report output format
- report screenshot retention policy
- report image format and quality
This is especially visible for long-running mobile automation sessions. Reports
may contain many high-resolution screenshots and grow into very large HTML files.
Although Core already supports html-and-external-assets, Studio does not expose
this capability.
The model environment configuration should remain responsible for model provider
environment variables. Runtime and report behavior should be configured through
a separate, versioned Studio advanced-settings model.
Initial scope:
-
Report output strategy:
single-htmlhtml-and-external-assets
-
Report screenshot retention:
- retain all screenshots
- retain no screenshots
- retain screenshots only for selected action types, such as
Tap
-
Report image persistence:
- preserve the original image
- encode report assets as JPEG
- configure JPEG quality
The configuration should apply consistently to Web, Android, iOS, HarmonyOS,
and Computer sessions, including Agent recreation.
What does the proposed API look like?
{
"schemaVersion": 1,
"agentOptions": {
"replanningCycleLimit": 10,
"waitAfterAction": 500,
"screenshotShrinkFactor": 2
},
"reportOptions": {
"outputFormat": "html-and-external-assets",
"screenshotFilter": {
"mode": "actions",
"actions": ["Tap"]
},
"imageFormat": "jpeg",
"imageQuality": 80
}
}
Contributor guide
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by tracing Studio's Agent creation and recreation paths for the Web, Android, iOS, HarmonyOS, and Computer runtimes, then compare them with the existing AgentOpt and Core report options. Done means a versioned advanced-settings model configures the listed agent and report behavior consistently across all sessions.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- typescript
- Domain
- testing, tooling
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 45/100