MAGIC Agent Skills is now open source! Star on GitHub
MAGIC Agent SkillsMAGIC Agent Skills
Commands

/magic:help

/magic:help — MAGIC Command Reference

Display this table when the user asks for help or available commands.

Available Commands

CommandWhat It DoesWhen to Use
/magic:lifecycleRun the full data pipeline (Discover → Plan → Execute → Validate → Deliver)Start or continue a phased data processing workflow
/magic:statusQuick snapshot of workspace stateSee current phase, quality score, recent actions
/magic:exploreEnter interactive data exploration modeInvestigate data freely before committing to a plan
/magic:findingsShow structured findings from profiling/analysisReview discovered issues as actionable proposals
/magic:proposeGenerate a processing plan from findingsGet recommendations for how to handle discovered issues
/magic:decideRecord a user decisionFormalize a choice (e.g., "go with option A")
/magic:specCreate or show the data processing specDefine schema, quality targets, and processing steps
/magic:reviewReview decisions and progressSee what's been decided and done so far
/magic:init-workspaceInitialize workspace directory structureStart a new data project with standardized layout
/magic:helpShow this command referenceWhen you need a reminder of available commands
/magic:rollbackRevert to a previous checkpointUndo a processing step that produced bad results
/magic:connectConnect to a database and inspect schemaSet up a database connection for data extraction
/magic:deliverDeliver processed data to DB, HuggingFace, or local fileExport final results to an external destination

Tip: You don't need to memorize these commands. Natural language works just as well — say "what's the quality score?" instead of /magic:status, or "help me clean this data" instead of remembering which skill to invoke.

Was this page helpful?
Edit on GitHub

Last updated on

On this page