explainlisted
Install: claude install-skill real-case/marvin-toolkit
Explain the selected code or the area the user is asking about.
## Approach
1. Read the code in question
2. Trace execution flow — follow calls, data transformations, state changes
3. Check for related tests to understand expected behavior and edge cases
4. If the code contains non-obvious decisions, check git blame/log for historical context
5. Identify the architectural pattern (if any) and its trade-offs in this context
## Explanation structure
Tailor depth to complexity. Simple utilities need 2–3 sentences. Complex systems need a full walkthrough.
**Always include:**
- **What it does** — one-sentence summary
- **How it works** — step-by-step walkthrough of the logic
**Include when relevant (skip if trivial):**
- **Why it's designed this way** — architectural rationale, constraints, trade-offs
- **Key dependencies** — what this code depends on and what depends on it
- **Gotchas** — non-obvious behavior, edge cases, known limitations, potential footguns
## Calibration rules
- Match the person's apparent expertise level — don't over-explain fundamentals to a senior engineer, don't assume domain knowledge from a newcomer
- Lead with the "what" before the "how" — the reader should know the purpose before diving into mechanics
- When multiple patterns or approaches coexist, explain *why* rather than just listing them
- For legacy or unusual code, prioritize explaining *intent* over *mechanics* — what was the author trying to achieve?
- If the code has bugs or anti-patte