Skill: cross-app-workflow-coordination

[Case 2]

• Initial Environment: A mobile device with Waze Navigation and Uber installed. The device is currently on the home screen.
• Real Question: Locate a nearby gym using Waze Navigation and then book a ride through Uber to get there.
• Real Trajectory: 1. Open Waze Navigation and search for 'gym'. 2. Select a nearby result, 'Trojan Crossfit'. 3. Observe the details and store 'Trojan Crossfit' into memory. 4. Exit Waze and open the Uber app. 5. In the Uber destination field, recall the name 'Trojan Crossfit' and input it.
• Real Answer: An Uber ride is initiated with 'Trojan Crossfit' as the destination.
• Why this demonstrates the capability: This interaction involves a 'Chain-of-Memory' where the output of a navigation search must be converted into a structured knowledge seed for a ride-hailing application, bridging two disjointed application environments.

[Case 3]

• Initial Environment: A web browser is open on a 'Startup Funding Application' page. The user has provided an external plan document containing the company's founding date, business stage, and funding amount.
• Real Question: Please fill out the startup funding application using the details in my company profile.
• Real Trajectory: 1. Scan the company profile for 'Business Stage'. 2. Switch focus to the browser, locate the 'Business Stage' dropdown on the form, and expand it to select 'Seed'. 3. Locate the 'Funding Amount' field and retrieve '500000' from memory to type. 4. Switch to date selection for founding date. 5. Verify all entered data against the document memory and submit.
• Real Answer: The funding application is successfully submitted with all fields populated from the document.
• Why this demonstrates the capability: This demonstrates multimodal info-transfer across applications because the agent must align semantic concepts from a distinct source document with heterogeneous, complex GUI widgets (dropdowns, inputs) on a destination web form, actively maintaining the memory abstraction across the workflow.

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

2. Phase 2: Trajectory Selection Once Phase 1 is complete, read the selection criteria to evaluate the trajectory: references/SELECTION.md

3. Phase 3: Data Synthesis Once a trajectory passes Phase 2, read the synthesis instructions to generate the final data: references/SYNTHESIS.md