CLI Overview - selftune

Usage

selftune [command] [options]

Running selftune with no arguments shows the status (same as selftune status).

Command	Description
`init`	Initialize agent identity and configuration
`doctor`	Health check: logs, hooks, config, permissions
`quickstart`	Guided onboarding walkthrough
`uninstall`	Clean removal of selftune

Command	Description
`sync`	Replay and normalize session data from all platforms
`ingest`	Ingest sessions from a specific platform

Command	Description
`status`	One-line health summary per skill
`last`	Quick insight from most recent session
`verify`	Draft-package verification alias that checks readiness, then emits measured report evidence
`grade`	Grade skill execution quality
`eval`	Generate, run, and manage evaluation sets

Command	Description
`create`	Initialize a draft skill package skeleton
`publish`	Draft-package publish alias with watch enabled by default
`workflows`	Scaffold new skills from repeated telemetry patterns

Command	Description
`improve`	Alias for the smallest matching evolution surface
`evolve`	Propose and deploy skill description improvements
`watch`	Monitor for post-deploy regressions
`run`	Alias for the autonomous orchestrate loop
`orchestrate`	Autonomous sync → grade → evolve → watch loop

Command	Description
`dashboard`	Open local web dashboard
`cron`	Set up automated scheduling
`contributions`	Manage community contributions
`badge`	Generate embeddable health badges
`export`	Export data for backup or analysis

selftune installs these Claude Code hooks automatically:

Hook	Event	Purpose
`prompt-log`	UserPromptSubmit	Log every user query
`auto-activate`	UserPromptSubmit	Evaluate activation rules
`skill-change-guard`	PreToolUse (Write/Edit)	Prevent unreviewed SKILL.md changes
`evolution-guard`	PreToolUse (Write/Edit)	Block conflicting edits during evolution
`skill-eval`	PostToolUse (Read)	Track skill triggers
`session-stop`	Stop	Capture session telemetry