`nervapack enrich`¶

Add semantic doc-to-code edges to an existing graph without full re-ingestion.

Synopsis¶

nervapack enrich [PATH] [OPTIONS]

Description¶

enrich reads the existing graph (built by nervapack ingest) and uses an LLM to create EXPLAINS edges between markdown documentation nodes and AST code nodes. This is useful when:

You ran ingest without an LLM (structural graph only) and now want to add semantic bindings.
You added new documentation and want to bind it to existing code without re-parsing the entire codebase.
You want to upgrade the binding quality (e.g., switch from keyword-based to LLM-based).

The command requires a running LLM (Ollama, Claude, or OpenAI). Unlike ingest, it will error if no LLM is available, since its sole purpose is semantic binding.

Options¶

Option	Description	Default
`PATH`	Path to the repository	`.` (current)
`--llm`	LLM provider (`ollama`, `claude`, `openai`, `mcp`)	Auto-detect
`--model`	Model name (provider-specific)	Provider default
`--api-key`	API key for cloud providers	From env vars

Examples¶

Enrich with local Ollama¶

nervapack enrich .
nervapack enrich . --llm ollama --model llama3

Enrich with Claude (shows cost estimate before proceeding)¶

nervapack enrich . --llm claude

Enrich a different repo¶

nervapack enrich /path/to/repo --llm openai --model gpt-4o

Expected Output¶

Enriching repository at .

Setting up LLM provider...
Using LLM provider: ollama (llama3:latest)

Adding semantic edges for 24 markdown chunks...
Enrichment complete. Added 87 semantic edges.

For cloud providers, a cost estimate is shown before binding starts:

💰 Cost Estimate
Provider: claude (claude-haiku-4-5-20251001)
Markdown chunks to bind: 24
Estimated cost: $0.03
Proceed with cloud LLM binding? [y/n]:

Notes¶

Existing EXPLAINS edges are preserved — only new edges are added (no duplicates).
If the graph doesn't exist yet, run nervapack ingest . first.
To re-bind from scratch, delete .nervapack/graph.graphml and run ingest again.

nervapack enrich¶