# Tool Mediated Information Acquisition

> Use this when the user wants planning data about looking things up before deciding, choosing the right search tool, or gathering scattered facts into one plan. Trigger it for requests like 'make planning tasks where the agent has to search first', 'give me itinerary problems that require checking flights and hotels', 'create planning tasks with missing information', or 'make questions where the plan depends on tool use instead of guessing.'

- Skill: `dingxingdi/tool-mediated-information-acquisition-2` (Agent Skill, multi-file: 5 files)
- Install (CLI): `npx skillmds@latest add dingxingdi/tool-mediated-information-acquisition-2`
- Raw SKILL.md: https://api.skillmd.com/api/skills/dingxingdi/tool-mediated-information-acquisition-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/tool-mediated-information-acquisition-2

---


# Skill: tool-mediated-information-acquisition
## 1. Capability Definition & Real Case
* **Professional Definition**: The ability to identify what external information is still missing, choose the right tools to retrieve it, and accumulate that information into a usable working context before finalizing a plan.
* **Dimension Hierarchy**: Open-World Real-World Planning->Information-Grounded Plan Construction->tool-mediated-information-acquisition
### Real Case
**[Case 1]**
* **Initial Environment**: A travel-planning agent receives a query for a five-day trip from Seattle to California with a budget cap, a room-type preference, and a pet-friendly lodging requirement. The agent starts with no itinerary details and has access to CitySearch, FlightSearch, DistanceMatrix, RestaurantSearch, AttractionSearch, AccommodationSearch, and a notebook for storing retrieved facts.
* **Real Question**: Create a five-day travel plan from Seattle that satisfies the budget and lodging preferences while choosing feasible cities, transportation, dining, attractions, and accommodations.
* **Real Trajectory**: Search candidate California cities. Probe flights for the departure date and rule out cities with no available route. Check distance and cost for inter-city transfers. Query accommodations that satisfy the room rule and room type, then gather restaurants and attractions for each chosen city and record all accepted results before drafting the final itinerary.
* **Real Answer**: A good answer first uses search tools to build an evidence-backed notebook, then produces a plan whose entries are drawn from retrieved flights, accommodations, restaurants, attractions, and transfer costs rather than invented details.
* **Why this demonstrates the capability**: This demonstrates tool-mediated acquisition because the agent cannot plan reliably from the user request alone. The right plan depends on selecting the correct retrieval tool at each step and preserving results in working memory. The benchmark therefore tests information gathering as a planning prerequisite, not as a cosmetic add-on.

## 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`