standards

standards answers the question “what’s our standard for X, and why?” from your team’s own written sources — and cites where the answer came from. It’s the read-only counterpart to writing a standard: you ask, it looks it up, it never makes something up.

What it solves

Conventions live scattered across ADRs, rules files, a vocabulary doc, and captured knowledge — so “do we use X or Y here?” gets answered from memory, or guessed, and the guess drifts from what the team actually decided. New people can’t self-serve; the answer with its reasoning is buried. This skill makes the team’s own decisions queryable, with the rationale attached.

How it works

• Precedence-ordered lookup — it consults sources in a fixed order: docs/adr/ (decisions) → rules/ (coding rules, including *.local.md overrides) → CONTEXT.md (vocabulary) → knowledge/ (confirmed facts). The highest-precedence source that covers the topic wins.
• Cited, with rationale — the answer names its source and explains why the standard is what it is, not just what it is.
• Never invents — when no source covers the topic, it says so plainly and routes you to author the missing standard (doc-adr, or a rules override) rather than fabricating one.
• Read-only, never gates — it informs; it never blocks a commit or a change.

It pairs with audit-grounding (which flags drift in a diff against the same sources) and doc-adr (which authors the standard that’s missing).

When to use

Any time you need to know the team’s standard on something and the reasoning behind it — before writing code that should conform, onboarding, or settling an “is this how we do it here?” question. Registered in the docs bundle.