Commands

The commands group is the new home for detailed command documentation. Each leaf page explains one executable CLI command from the user point of view and the agent integration point of view.

Rendering diagram...

Start Here

• Read init when you are setting up RAGit in a repository.
• Read log when you want to understand how snapshot-scoped knowledge, beliefs, and open loops changed across commits.
• Read timeline when you want the operational time axis for session, memory, harness, and ingest events.
• Read drift when you want to know which knowledge objects should be revalidated at the current HEAD.
• Read repair when you want a conservative action plan, or safe automatic execution, from current drift results.
• Read security audit when you want to inspect whether secret-like content or unsafe remote egress posture leaked into RAGit-owned state.
• Read query when you want direct retrieval hits from indexed knowledge.
• Read context pack when you want a compact prompt packet for the next agent step.
• Read memory recall when you want to resume active work instead of doing a general search.
• Read session materialize when you want to extract session artifacts from transcripts and tool traces.
• Read harness run when you want to execute a stored harness suite and collect deterministic evaluation results.

Command Families

• Top-level commands: init, log, timeline, drift, repair, describe, ingest, query, status, doctor
• Doc: doc
• Config: config
• Context: context
• Hooks: hooks
• Security: security
• Session: session
• Artifact: artifact
• Harness: harness
• Memory: memory
• Migrate: migrate

How To Read These Pages

• Every leaf page uses the same structure so you can find syntax, options, payload examples, and cautions quickly.
• Commands that support --input <path|-> include a JSON payload example.
• Commands that support --view explain the difference between minimal, default, and full.
• Commands that support --dry-run explain what is validated before any real mutation happens.

Fast Mappings

• query answers: "What indexed knowledge matches this question?"
• log answers: "What changed semantically across snapshot history, and what should the next agent still believe?"
• timeline answers: "What happened across session, memory, harness, and ingest workflows?"
• drift answers: "Which knowledge objects are still trustworthy at the current HEAD, and which ones need review?"
• repair answers: "Given current drift, what conservative repair actions should I plan or safely apply?"
• security audit answers: "Has raw-looking sensitive content or unsafe remote egress posture leaked into RAGit-owned state?"
• context pack answers: "What is the smallest useful prompt packet for this goal?"
• memory wrap answers: "What should I record before stopping?"
• memory recall answers: "What should I load before resuming?"
• memory promote answers: "What should become durable project knowledge?"

If you still prefer the old entry point, the legacy CLI Reference page now redirects you into this group.