# Why

> Reconstruct why code or a product decision exists from source history, issues, documents, conversations, observability, and analytics. Use for $why, design rationale, regressions, postmortems, or data-backed thresholds. Use $how for execution behavior.

- Skill: `vaibtan/why` (Agent Skill, multi-file: 12 files)
- Install (CLI): `npx skillmds@latest add vaibtan/why`
- Raw SKILL.md: https://api.skillmd.com/api/skills/vaibtan/why/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Data & Analytics
- Author: Vaibtan (https://skillmd.com/u/vaibtan)
- Updated: 2026-09-22
- Page: https://skillmd.com/skills/vaibtan/why

---


# Why it exists

Treat rationale as an evidence question. Current code explains what exists; history and operational records may explain why.

1. State the target and your interpretation when it is ambiguous.
2. Inventory evidence available in the current Codex environment. Source control is usually local. Other categories may be available through installed apps, MCP tools, or files the user placed in scope.
3. Search only relevant categories. Use [source playbook](references/source-playbook.md) and one matching file under `references/sources/` when a connector is available.
4. For broad evidence sets, give one read-only investigator each to source history, issue tracker, long-form docs, team chat, observability, error tracking, or analytics. Do not ask one worker to query unrelated systems.
5. Synthesize with [epistemic rules](references/epistemics.md). Corroborate claims, preserve dates and links, and label inference or missing coverage.

Do not claim that an unavailable source was searched. Authentication failures and absent connectors are coverage gaps. Never fabricate rationale from current code shape.

Return the current answer first, then a compact source-and-coverage note. If the investigation precedes a change, finish with Preserve, Change, Avoid, and Risk constraints grounded in the evidence.