task-inquiry

Clarify consequential uncertainty through evidence, reasoning, and useful questions.

Task Inquiry helps an assistant resolve uncertainty that can change what a task should do next. It begins with a real mismatch or tension, develops plausible explanations, and uses evidence or a focused user question to distinguish them. The assistant carries the investigation and execution forward instead of requiring the user to formalize the whole problem.

Use this Skill

Make the complete task-inquiry/ directory available through the Skill mechanism supported by your assistant, then invoke Task Inquiry when a material ambiguity, conflicting explanation, or unresolved choice affects the task. The package contains the standalone instructions in SKILL.md. It has no required companion framework, private context, scripts, or provider integration.

This method is useful when inquiry can improve the result. A clear, settled task should proceed directly. A question is useful only when its answer can change the interpretation, evidence, or action.

Examples

Investigate an unexpected result. A report's totals differ from the dashboard. The assistant checks date ranges, filters, and definitions, identifies which differences could explain the mismatch, then asks about the intended reporting period only if the available evidence cannot determine it.

Resolve a meaningful preference. Two proposed plans meet the stated goal, but one saves time while the other preserves flexibility. The assistant explains that tradeoff and asks which outcome matters more before taking the dependent step; it continues any independent research meanwhile.

Method at a glance

The assistant identifies the actual tension, distinguishes facts from assumptions and unknowns, and proposes explanations worth checking. It derives what would differ under each explanation, gathers available evidence, and updates its judgment. In ordinary work, it asks the user for information or a decision the evidence cannot supply when that answer would change the work. Teaching may also use questions to reveal or develop understanding, as described below. The host's own interaction and authorization boundaries remain in force.

Work and learning

In ordinary work, useful feedback can help the user notice a relationship: show what the explanation predicted and what the evidence showed. The assistant still carries the reasoning and execution, without adding lessons or exercises. In teaching, questions and practice may be chosen to reveal or develop the learner's understanding even when the answer is known. The teaching role chooses the method and owns progression; inquiry does not impose a teaching format on ordinary tasks.

License

This independent package uses the MIT License. You may obtain the complete directory from GitHub and use, modify, and redistribute it, including commercially, with the required MIT notices. It does not include Yinyang. The public collection provides downloads and Issue discussion; this collection does not accept pull requests.