# Task Progress Decision Making

> Use this skill when the user wants questions like 'what should happen next', 'which action should the agent take now', 'based on the progress so far', or 'make the video behave like a step-by-step task.' Trigger it for clips with goals, partial completion, and candidate actions where the right choice depends on temporal progress and current state.

- Skill: `dingxingdi/task-progress-decision-making-2` (Agent Skill, multi-file: 5 files)
- Install (CLI): `npx skillmds@latest add dingxingdi/task-progress-decision-making-2`
- Raw SKILL.md: https://api.skillmd.com/api/skills/dingxingdi/task-progress-decision-making-2/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: dingxingdi (https://skillmd.com/u/dingxingdi)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/dingxingdi/task-progress-decision-making-2

---


# Skill: task-progress-decision-making
## 1. Capability Definition & Real Case
* **Professional Definition**: The capability to infer the correct next action from a temporally grounded task trajectory, current observation, and task goal, especially when the answer depends on progress state rather than static scene recognition alone.
* **Dimension Hierarchy**: Long-Horizon Reasoning->Inference and Planning->task-progress-decision-making
### Real Case

**[Case 1]**
* **Initial Environment**: A first-person household-cleaning video paired with a current observation image at the sink. The task goal is explicitly to clean the sink, and the clip shows the cloth has already been rinsed.
* **Real Question**: Considering the progress shown in the video and my current observation shown in the image, what action should I take next in order to clean the sink?
* **Real Trajectory**: Recover the recent task history from the video, verify that rinsing has already occurred, inspect the current image to confirm the tap is still running and the sink area is ready for the next step, then choose the next action that advances the goal rather than repeating a completed step.
* **Real Answer**: Close tap.
* **Why this demonstrates the capability**: The problem is about progress-aware decision making: the same sink image could support several actions unless the preceding task trajectory is known. The correct answer therefore depends on aligning task goal, completed steps, and current state into one action choice.

---
## 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`