dingxingdi/paper_fast_search_backup/cross-app-workflow-coordination

Skill: cross-app-workflow-coordination

1. Capability Definition & Real Case

• Professional Definition: The capability to decompose a user goal into app-specific subtasks, execute them in the right order across multiple applications, and preserve the necessary information between transitions. The agent must coordinate scheduling, execution, and cross-app information flow so that later subtasks receive the correct outputs of earlier ones.
• Dimension Hierarchy: Goal-Directed GUI Workflow Execution->Verified Task Completion->cross-app-workflow-coordination

Real Case

[Case 1]

• Initial Environment: A mobile phone contains a travel-related information source, a notes application, and a clock application. The task requires the agent to gather flight details, preserve the relevant arrival-time information, and then use that information in a different app to set an alarm.
• Real Question: Create a note with the information of an one-way flight from Shanghai to London, then set the alarm for the arrival time of the flight.
• Real Trajectory: Retrieve the flight details from the relevant source, create a note containing the route and arrival-time information, switch to the clock app, and set the alarm using the transferred arrival time.
• Real Answer: A note containing the flight information is saved, and an alarm is set to the flight’s arrival time.
• Why this demonstrates the capability: No single app contains the full solution state. The agent must preserve the right piece of information across app boundaries and avoid losing or corrupting it during transitions. That makes the task a direct test of cross-app scheduling and information hand-off.

Pipeline Execution Instructions

To synthesize data for this capability, you must strictly follow a 3-phase pipeline. Do not hallucinate steps. Read the corresponding reference file for each phase sequentially:

1. Phase 1: Environment Exploration

Read the exploration guidelines to discover raw knowledge seeds: references/EXPLORATION.md

1. Phase 2: Trajectory Selection

Once Phase 1 is complete, read the selection criteria to evaluate the trajectory: references/SELECTION.md

1. Phase 3: Data Synthesis

Once a trajectory passes Phase 2, read the synthesis instructions to generate the final data: references/SYNTHESIS.md