| name | ai-explain |
| description | Explains code, concepts, patterns, and architecture with engineer-grade depth: 3-tier control (brief/standard/deep), ASCII diagrams, execution traces, anchored to real file:line references. Trigger for 'how does this work', 'why does it do X', 'trace through this', 'explain this pattern', 'walk me through'. Not for documentation artifacts; use /ai-prose or /ai-docs instead. Not for fixing code; use /ai-debug or /ai-build instead. |
| effort | mid |
| argument-hint | [topic]|--depth brief|standard|deep |
| tags | ["explanation","teaching","analysis","architecture"] |
| model_tier | sonnet |
| mirror_family | antigravity-skills |
| generated_by | ai-eng sync |
| canonical_source | .claude/skills/ai-explain/SKILL.md |
| edit_policy | generated-do-not-edit |
Explain
Quick start
/ai-explain "how does the audit chain work" --depth brief
/ai-explain auth-handler.ts:42-90 --depth standard
/ai-explain "spec lifecycle FSM" --depth deep
Workflow
Read-only; §10.7 Clean Code — clarity over cleverness.
- Identify subject — classify into Code / Concept / Pattern /
Architecture / Error / Difference. If ambiguous, ask ONE question.
- Search codebase — search the tree for real instances;
file:line refs
are primary evidence. If none, use a generic stack example and say so.
- Select depth (per cue table) and deliver only that depth's sections.
- Follow-up — "what about X?" extend at depth; "go deeper" +1 level,
only new sections; "trace a different path" re-run Trace; "show me in
my code" search the codebase for the concept.
| Depth | Cues | Sections |
|---|
| Brief | "TL;DR", "brief", "short" | Summary + Walkthrough (3-5 steps) |
| Standard (default) | general question | + Diagram + Gotchas + Trace |
| Deep | "deep dive", "everything", "teach me" | + Context Map + Complexity Notes |
Section contract:
- Summary — 1-2 technical sentences: what it does + why it exists.
- Walkthrough — numbered
file:line steps in execution order.
- Diagram — ASCII (data flow / call chain / state machine / sequence),
width < 70 chars, no Mermaid.
- Gotchas — specific edge cases / perf traps / concurrency hazards in
THIS code, not generic advice.
- Trace It — execution trace of a concrete scenario; show data
transformation per step, highlight decision points.
- Context Map — when to use / when NOT / alternatives with tradeoffs.
- Complexity Notes — cyclomatic complexity, nesting, time/space of hot
paths, concurrency assessment.
Pitfalls: don't over-explain standard concepts (assume competence); no
generic gotchas; avoid "basically" / "simply" / "just".
Examples
Example — quick orientation on an unfamiliar pattern
User: "how does the audit chain work in this repo?"
/ai-explain "audit chain" --depth brief
Anchors to framework-events.ndjson; emits a 3-paragraph overview with a
small ASCII flow.
Integration
Called by: user directly, /ai-onboard (teaching mode). Read-only. See
also: /ai-debug, /ai-prose content blog, /ai-verify (architecture
assessment).
$ARGUMENTS
