Skip to content

CLI Reference: Init, Scan & Enrich

Commands for initializing and updating the repository graph.


tldrgraph init

Initializes or refreshes the multi-layer graph, discovers layers, and generates embeddings.

tldrgraph init [OPTIONS]

Options

Flag Type Default Description
--batch integer 200 Chunk size for LLM enrichment batching.
--limit integer None Caps total enriched nodes in this run.
--embeddings auto / off auto Toggle dense FastEmbed ONNX embedding generation.
--yes, -y boolean False Assume yes to confirmation prompts.
--json boolean False Output status as structured JSON.

Aliases

tldrgraph scan and tldrgraph enrich are aliases for tldrgraph init, maintained for backward compatibility.


tldrgraph install

Installs coding agent slash commands, skills, and execution rules across your local tools.

tldrgraph install [OPTIONS]

Options

Flag Type Default Description
--all-agents boolean False Install workflows for all supported agents regardless of project markers.
--tool string None Target a specific agent (e.g. claude, cursor, codex, windsurf).

Resumable States

When running in automated agent workflows, tldrgraph init reports resumable execution states:

Status Meaning Action Required
needs_layers Layers are unconfigured. Agent reads code and designs .tldrgraph/layers.config.yaml.
needs_confirmation LLM token spend confirmation. Agent prompts user once before enrichment batch.
needs_enrichment Nodes awaiting semantic summary. Agent opens source files and adds architectural intent.
needs_embeddings Embeddings not yet compiled. Downloads ONNX model and indexes vector store.
done Initialization complete. Graph is ready for querying and visualization.