Skip to content

chore(o-gui-agent): optimize agent game mode's processing #1664

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ZhaoHeh merged 2 commits into from
Sep 26, 2025
Merged

chore(o-gui-agent): optimize agent game mode's processing #1664

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3078654
feat(o-gui-agent): comment out prompts for game mode
ZhaoHeh Sep 26, 2025
03c7b64
fix(o-gui-agent): fix bugs of inserting preset messages on game mode
ZhaoHeh Sep 26, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
21 changes: 11 additions & 10 deletions multimodal/omni-tars/gui-agent/src/GuiAgentPlugin.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -25,6 +25,7 @@ export class GuiAgentPlugin extends AgentPlugin {
readonly environmentSection = COMPUTER_USE_ENVIRONMENT;
private agentMode?: AgentMode;
private operatorManager: OperatorManager;
private initializedWithMode = false;

constructor(option: GuiAgentPluginOption) {
super();
Expand All @@ -36,10 +37,11 @@ export class GuiAgentPlugin extends AgentPlugin {
}

async initialize(): Promise<void> {
if (this.agentMode && this.agentMode.id === 'game') {
if (this.agentMode && this.agentMode.id === 'game' && !this.initializedWithMode) {
await this.emitPresetUserQuey();
await this.emitScreenshotEvent();
await this.emitPresetAssistantMessage();
this.initializedWithMode = true;
}

this.agent.registerTool(
Expand Down Expand Up @@ -109,7 +111,7 @@ export class GuiAgentPlugin extends AgentPlugin {
if (interval > 0) await sleep(interval);
const output = await operator?.doScreenshot();
if (!output?.base64) {
guiLogger.error('Failed to get screenshot');
guiLogger.error('emitScreenshotEvent: Failed to get screenshot');
return;
}
const base64Tool = new Base64ImageParser(output.base64);
Expand All @@ -126,7 +128,7 @@ export class GuiAgentPlugin extends AgentPlugin {
const compressedSize = compressedBuffer.byteLength;
const compressionRatio = (((originalSize - compressedSize) / originalSize) * 100).toFixed(2);

guiLogger.debug(`compression stat: `, {
guiLogger.debug(`emitScreenshotEvent compression stat: `, {
originalSize: formatBytes(originalSize),
compressedSize: formatBytes(compressedSize),
compressionRatio: `${compressionRatio}% reduction`,
Expand All @@ -150,7 +152,7 @@ export class GuiAgentPlugin extends AgentPlugin {

const eventStream = this.agent.getEventStream();
const events = eventStream.getEvents();
guiLogger.info('onAfterToolCall events length:', events.length);
guiLogger.info('emitScreenshotEvent events length:', events.length);

const event = eventStream.createEvent('environment_input', {
description: 'Browser Screenshot',
Expand All @@ -177,11 +179,10 @@ export class GuiAgentPlugin extends AgentPlugin {
private async emitPresetAssistantMessage(): Promise<void> {
const eventStream = this.agent.getEventStream();
const events = eventStream.getEvents();
if (events.length === 0) {
const event = eventStream.createEvent('assistant_message', {
content: `Successfully navigated to ${this.agentMode!.link}, page loaded successfully`,
});
eventStream.sendEvent(event);
}
guiLogger.debug('emitPresetAssistantMessage events length:', events.length);
const event = eventStream.createEvent('assistant_message', {
content: `Successfully navigated to ${this.agentMode!.link}, page loaded completely`,
});
eventStream.sendEvent(event);
}
}
10 changes: 5 additions & 5 deletions multimodal/omni-tars/omni-agent/src/index.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -35,7 +35,7 @@ export default class OmniTARSAgent extends ComposableAgent {
{
title: 'Four in a Row',
category: 'Game',
prompt: `Play this game, win the computer`,
// prompt: `Play this game, win the computer`,
image:
'https://img.poki-cdn.com/cdn-cgi/image/q=78,scq=50,width=80,height=80,fit=cover,f=auto/e80686db-b0fb-4f2c-bd2f-3a89734f102a/four-in-a-row.jpg',
agentOptions: {
Expand All @@ -49,7 +49,7 @@ export default class OmniTARSAgent extends ComposableAgent {
{
title: 'Block the Pig',
category: 'Game',
prompt: `Play this game, reach level 5`,
// prompt: `Play this game, reach level 5`,
image:
'https://img.poki-cdn.com/cdn-cgi/image/q=78,scq=50,width=80,height=80,fit=cover,f=auto/9fec1234ce2afd5e789f56da463dcffc/block-the-pig.jpeg',
agentOptions: {
Expand All @@ -63,7 +63,7 @@ export default class OmniTARSAgent extends ComposableAgent {
{
title: 'Factory Balls Forever',
category: 'Game',
prompt: `Play this game, reach level 5`,
// prompt: `Play this game, reach level 5`,
image:
'https://img.poki-cdn.com/cdn-cgi/image/q=78,scq=50,width=80,height=80,fit=cover,f=auto/2a503d0a1d9475d6e62c7ea11caa429ab952aa8f500755613a34e66e2196fe82/factory-balls-forever.png',
agentOptions: {
Expand All @@ -77,7 +77,7 @@ export default class OmniTARSAgent extends ComposableAgent {
{
title: 'Snake Solver',
category: 'Game',
prompt: `Play this game, reach level 5`,
// prompt: `Play this game, reach level 5`,
image:
'https://img.poki-cdn.com/cdn-cgi/image/q=78,scq=50,width=80,height=80,fit=cover,f=auto/e466baf845544ef47e0dfbd60f127071/snake-solver.png',
agentOptions: {
Expand All @@ -91,7 +91,7 @@ export default class OmniTARSAgent extends ComposableAgent {
{
title: 'Penalty Kicks',
category: 'Game',
prompt: `Play this game, your target score is 600`,
// prompt: `Play this game, your target score is 600`,
image:
'https://img.poki-cdn.com/cdn-cgi/image/q=78,scq=50,width=80,height=80,fit=cover,f=auto/0770daaa8c4ff3c36dd53e6e41f59396/penalty-kicks.png',
agentOptions: {
Expand Down
Loading