hotmess

name

hotmess

description

Find and explain churn hotspots and unstable code clusters in a git repo. Goes beyond a flat "most-edited files" list — auto-detects clusters by directory, by filename stem (foo.go + foo_test.go), and by co-change (files that change together) — then classifies each cluster (unstable / under-development / buggy / tightly-coupled / spec-shifting) and explains *why* it's a problem. Trigger whenever the user asks about churn, hotspots, unstable code, files that change a lot, refactoring candidates, technical debt hotspots, "what's a mess", "/hotmess", or any combination of folder + time window where they want insight rather than just numbers. Auto-detects code files when no extension is given.

hotmess

Identifies clusters of high-churn code and explains why they're problematic — not just a ranked file list. The user wants insight: which areas of the codebase are unstable, why, and what to do.

When to use

Trigger phrases: "hotmess", "hotspots", "what's churning", "unstable code", "where's the technical debt", "what's a mess in X", "/hotmess ". Always invoke for these — even if the user gives no time window or extension, fall back to sensible defaults (30 days, auto code detection).

How to run

Always invoke this skill's bundled analyzer (scripts/analyze.py, in this skill's base directory), which emits structured JSON with four signal sets. Do not reimplement git log parsing inline.

"<SKILL_DIR>/scripts/analyze.py" <path> --since "<expr>" [--ext .go,.ts] [--top N] [--html /tmp/hotmess.html] [--quiet]

<SKILL_DIR> is the "Base directory for this skill" path provided when the skill is invoked. Run from inside the target git repo (or pass an absolute path; git resolves it).

Defaults: --since "30 days ago", --top 30, auto-detected code extensions.

If the user asks for a visualization, diagram, picture, "show me", chart, or graph — also pass and then . Use together with if you don't need the JSON in the same call. The HTML page contains a Mermaid module map (modules colored by classification, edges = cross-directory co-change), plus collapsible tables for stems and co-change pairs.