Command Map

Command Map

Home | Quick Start

Think of skm as four shelves: inspect, choose, visualize, and clean.

Inspect

Command	Use it when	Output
skm doctor	You want to check the local environment	Node version, directories, catalog, advisor CLI, CI note
skm scan	You need fresh inventory facts	Catalog, counts, categories, MCP summary
skm / skm status	You want a one-screen health check	Score, risks, suggested next commands
skm risks	You want a risk list before touching anything	Prioritized risk items and conservative suggestions
skm sessions	Session logs feel too large	Workspace-level log distribution

Choose

Command	Use it when	Output
skm search <text>	You remember a keyword	Ranked search results
skm ask "task"	You want the best skill quickly	Best match, reason, alternatives
skm recommend "task" --why	You want to compare candidates	Ranked table with score and matched reasons

Visualize

Command	Use it when	Output
skm report	You want one local health summary	Terminal summary, JSON, or HTML report
skm graph	You want relationships	Summary, JSON, Mermaid, or interactive HTML graph
skm dupes	You want duplicate details	Same name, same content, same category, similar text
skm audit	You want real usage facts	Top used, zombie skills, MCP usage, context cost

Clean, Carefully

Command	What it changes	Guardrails
skm sessions --clean	Deletes session logs	Requires policy, prints plan, confirmation, 24-hour safety window
skm disable <skill>	Renames skill directories	Reversible, no deletion, refuses plugin skills
skm enable <skill>	Restores renamed skill directories	Does not overwrite existing target directories
skm disable --mcp <name>	Edits MCP config	Backup and confirmation
skm enable --mcp <name>	Restores MCP config	Backup and no-overwrite behavior

Script Mode

Most read commands support --json:

skm scan --json
skm risks --json
skm recommend "markdown to html" --json
skm sessions --json

JSON field names remain stable across language settings.