spec-explain

Synthesizes feature documentation from specs, plans, changelogs, and ADRs into visual explanations.

Updated Mar 11, 2026
One-click install
npx skills add https://github.com/julien-m/livespec --skill spec-explain-julien-m
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: spec-explain
Source: https://github.com/julien-m/livespec/tree/main/.agent-sync/skills/spec-explain
Command: npx skills add https://github.com/julien-m/livespec --skill spec-explain-julien-m

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Specs and code drift apart over time, and months later nobody remembers why a feature was built the way it was. This Skill answers "how does X work?" by synthesizing information from spec files, implementation maps, changelogs, and architecture decision records without reading source code. ## Core Features & Use Cases - Living Documentation: Generates an 8-section structured explanation (what it does, who uses it, flows, data model, file locations, rationale, history) with embedded Mermaid diagrams. - Intent-Based Query Resolution: Classifies questions as how/why/what-changed/where and prioritizes the right sources (ADRs for why, changelog for history, implementation map for locations). - Natural Language Search: Accepts feature names or free-form questions, ranking the top 3 candidate features when multiple matches exist. - Use Case: A new team member asks "why did we choose Supabase over Firebase?" and receives an ADR-grounded answer with traceability references in seconds. ## Quick Start Ask the assistant to run /spec-explain with a feature name like "notifications" or a natural language question such as "how does the real-time messaging system work?"

Frequently Asked Questions about spec-explain

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

FAQPage Schema
How do I explain what a feature does without reading the code?

Run /spec-explain with the feature name or a natural language question. It reads the feature's spec.md, plan.md, implementation.md, and changelog.md, then produces a structured summary with Mermaid diagrams covering behavior, flows, data model, and file locations.

How to find out why a technology decision was made in a project?

Ask a why-question like "why did we choose Supabase over Firebase?" The command classifies the intent as ADR-first and searches .specs/stacks/decisions/ for the relevant architecture decision records, returning the documented rationale.

Can I query feature history or recent changes with natural language?

Yes. Temporal questions like "what changed in auth last month?" are classified as changelog-first queries. The command searches .specs/changelog.md and the feature changelog, and the --since flag filters entries after a specific date.

What happens when multiple features match my question?

The command returns the top 3 candidate features with a one-line rationale for each and asks you to pick one. If you do not choose, it explains the best match and labels the output as an assumed feature.

Does spec-explain modify any project files?

No. It is a read-only command. Its expectations contract forbids writes to src/ and .specs/, expects no git changes, and produces only stdout Markdown output.