cxavireh/hotz-skills/read-paper

Read Paper

Deep-dive a single paper with interactive Q&A. Outputs a structured analysis, then lets the user ask follow-up questions.

Instructions

When this skill is invoked with <input>:

Step 1: Parse input and fetch paper content

Determine input type and fetch accordingly:

If fetching fails entirely, tell the user and suggest an alternative input method.

Step 2: Gather research context

Read the following files if they exist (skip silently if not found):

• docs/papers/landscape.md — user's literature map
• exp/summary.md — user's experiment history

Step 3: Delegate to @domain-expert

Invoke @domain-expert with this prompt:

Mode 4: Paper Deep-Dive

Paper content:

{fetched paper content}

Research context:

- Literature landscape: {landscape.md content or "not available"}

- Experiment history: {exp/summary.md content or "not available"}

Provide your structured deep-dive analysis (Methodology Skeleton, Assumptions & Limitations, Bridge Analysis), then enter interactive Q&A mode.

The user will then interact directly with domain-expert for follow-up questions. When the user says "save" / "archive" / "store" / "存档" / "保存" (with optional "as {name}"), domain-expert handles archival via its Mode 4 archive sub-flow.