Explicit Hard Constraint Satisfaction

Skill: explicit-hard-constraint-satisfaction

1. Capability Definition & Real Case

• Professional Definition: The ability to generate plans that satisfy explicitly stated user requirements, especially numeric limits and categorical preferences, without violating any of the required constraints while optimizing or organizing the rest of the plan.
• Dimension Hierarchy: Open-World Real-World Planning->Constraint Compliance->explicit-hard-constraint-satisfaction

Real Case

[Case 1]

• Initial Environment: A travel query specifies a fixed budget, a required room type, a room rule, a cuisine preference, and sometimes a transportation prohibition such as no self-driving. The agent can only choose among retrieved options that expose these fields explicitly.
• Real Question: Plan a five-day trip for two travelers with a budget of about $2,000, requiring both American and Mediterranean food, pet-compatible entire-room lodging, and no option that breaks the stated travel preferences.
• Real Trajectory: Collect candidate flights or ground transport, then screen lodging by room type and room rule, filter restaurants by cuisine, aggregate total expected cost, and discard any branch whose cumulative price or category labels violate the user request before finalizing the itinerary.
• Real Answer: A correct answer is an itinerary whose transportation, dining, lodging, and total estimated cost all remain within the explicitly stated requirement set.
• Why this demonstrates the capability: The challenge lies in conjunctive constraint satisfaction rather than in finding attractive options. The planner must respect every hard requirement simultaneously, even when some individually plausible choices would improve convenience but violate one categorical condition. This makes the task a direct probe of user-aligned planning under explicit rules.

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