# Why

> Use when asking why code or a design exists this way — design rationale, rejected alternatives, regression history, postmortems, or where a threshold came from — produces a cited confidence-graded note. Not for how the runtime flows (tour-system / scan) or library API facts (research).

- Skill: `jayden-dang/why` (Agent Skill, multi-file: 4 files)
- Install (CLI): `npx skillmds@latest add jayden-dang/why`
- Raw SKILL.md: https://api.skillmd.com/api/skills/jayden-dang/why/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Product & Planning
- Author: jayden-dang (https://skillmd.com/u/jayden-dang)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/jayden-dang/why

---


# Why

Answer **why it was shaped this way**, not what it does. Evidence before narrative. Code is not intent.

<HARD-GATE>
OVERRIDE SHIP (2026-09-08): same-day baseline for "code is not intent" already
complied 2/2. User override required this skill anyway.
</HARD-GATE>

Companion: `/tour-system` or a scan for mechanics; `research` for library/RFC facts.
This skill owns historical forcing functions.

## Iron Law

```
NO INTENT CLAIM WITHOUT A CITATION OR AN EXPLICIT INFERENCE LABEL
NO SHORTCUT BY READING CODE SHAPE AS MOTIVATION
```

## Steps

1. **Anchor.** Name target files/symbols. Seed with `git blame` / `git log --follow -p` / `gh pr view` for recent touches. Pass seeds to investigators.
2. **Coverage map.** List available tools/MCPs. Map each to a category: source control (always), issue tracker, long-form docs, team chat, infra observability, error tracking, product analytics. Spawn **one investigator per available category** in parallel (readonly tools ok; do not invent MCP access). Null result = finding. Skip only with a written reason (no MCP, or provably irrelevant).
3. **Synthesize.** Separate investigators from the synthesizer. Load `references/epistemics.md` beside this file and follow its tiers. Output sections in order:
   - The Question
   - The Code in Question
   - What We Found (direct evidence, cited)
   - What We Can Reasonably Infer (hedged, chain explicit)
   - Competing Hypotheses (if needed)
   - What We Don't Know (searches that returned empty)
   - Sources Consulted (one line per category, including nulls and skips)
4. **Present.** Do not flatten hedges into confident prose.

WHEN the `why` is a precursor to changing the code, append Preserve / Change / Avoid / Risk constraints for planning.

## Rationalizations

| Thought | Reality |
|---|---|
| "The null check proves they feared nulls" | Mechanics ≠ motivation. Find the ticket, PR, or incident |
| "One category is enough" | Default is coverage; nulls teach how the decision was made |
| "research already covers this" | research owns external facts; this owns design archaeology |
| "Baseline already complied — skip" | OVERRIDE: user required the skill |

## Red Flags

- Confident "because" with no citation
- Citing the code as proof of its own intent
- Skipping investigators by anticipation
- Empty "What We Don't Know"

## Done when

The note separates found / inferred / unknown, cites sources, and lists every category searched or skipped with reason.