Yitang OSCAR Research

Use this skill to turn loose research requests into decision-support work with explicit scope, owners, acquisition paths, and conclusions.

Core Rules

• Research exists to support a decision, not to collect information.
• Write the decision question before gathering material.
• Narrow scope aggressively; refuse “everything about this industry”.
• Require a named owner and a conclusion format for every research task.
• End with 该做 / 怎么做 / 不该做什么, not an information dump.

OSCAR Workflow

1. Objective

• State the decision that the research must support.
• Make the question concrete enough to be answered with evidence.

2. Scope

• Define what to include and what to exclude.
• Cut future possibilities, adjacent topics, and nice-to-have research.

3. Checklist

• List the exact objects, people, products, datasets, or documents to inspect.
• Prefer a finite list over generic “do some market research”.

4. Acquisition

• Specify how the information will be obtained: documents, interviews, tests, field checks, trials.
• Choose channels proportional to the importance of the decision.

5. Reasoning

• Synthesize findings back to the original decision.
• Write conclusion, next action, and explicit non-actions.

Default Output

Unless the user asks for another format, output these sections:

1. 调研目标
2. 调研范围
3. 调研对象清单
4. 信息获取路径
5. 主责 / 协同 / 截止时间
6. 结论需要回答的问题
7. 不做清单

Research Management Rules

• Do not run more research tracks in parallel than the project can absorb.
• Force each track to report against its objective, not against effort spent.
• Use short daily updates and a midweek calibration when multiple people are involved.

References

• Read references/framework.md for the full OSCAR method, templates, and anti-patterns.
• Read references/templates.md when the user wants a ready-to-fill research brief, task assignment sheet, interview guide, or conclusion summary.