why

Investigate design rationale behind code by searching historical evidence sources in parallel.

136|8|Updated May 9, 2026
One-click install
npx skills add https://github.com/Sma1lboy/rove --skill why-sma1lboy
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: why
Source: https://github.com/Sma1lboy/rove/tree/main/.agents/skills/pstack/skills/why
Command: npx skills add https://github.com/Sma1lboy/rove --skill why-sma1lboy

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Code tells you what it does but rarely why it exists. This Skill answers "why was this built this way" questions by fanning out parallel investigators across every available evidence source — git history, PRs, ADRs, issue trackers, docs, chat, observability, error tracking, and analytics — then synthesizing a confidence-calibrated, citation-backed answer instead of a confident guess. ## Core Features & Use Cases - Parallel evidence investigation: Spawns one investigator subagent per evidence category (source control, tickets, docs, chat, infra, errors, analytics, plus this repo's local ADRs and design notes), each with a source-specific playbook. - Confidence-tiered synthesis: A dedicated synthesizer classifies every claim as Direct, Supported, Inferred, Speculative, or Unknown, with mandatory citations and hedged language for inferences. - Honest gap reporting: Null results and unsearched sources are documented in a Sources Consulted coverage map rather than papered over. - Use Case: Ask "why is this retry limit set to 3?" and get back the PR discussion, linked ticket, and monitoring data that motivated it — or an explicit statement that no record exists. ## Quick Start Ask the agent "why was this code designed this way?" while pointing at a file, function, or constant in your repository.

Frequently Asked Questions about why

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I find out why code was written a certain way?▼

Ask a "why" question about a file, function, or constant. The skill anchors the target with git blame and log, spawns one investigator per available evidence source, and returns a cited answer separating direct evidence from inference.

What is the difference between the why and how skills?▼

The how skill explains what code does and how it works at runtime. The why skill explains the forces that shaped it: design rationale, rejected alternatives, business constraints, and incident history.

Does this work without MCP servers like Linear or Slack?▼

Yes. Git and GitHub are always searched, and this repo adds local sources like ADRs, docs/design notes, the daemon issue store, and the changelog. Missing MCP categories are reported as explicit gaps, not silently skipped.

Can it tell where a magic number or threshold came from?▼

It searches commit history, PR discussions, monitor thresholds, and analytics distributions for the value's origin. If no record exists, it says so specifically rather than inventing a rationale.

What are the limitations of historical code investigation?▼

Evidence is fragmentary: tickets go stale, chat history may be deleted, and squash merges collapse discussion. The skill mitigates this with confidence tiers and a mandatory What We Don't Know section, but some questions remain unanswerable from records alone.