How Flowgraph works

Flowgraph (npm: @igentify/helix-flowgraph) is a static analysis tool: it turns source into a versioned technical graph, then layers product features and optional AI enrichments. Queries read scans/<scanId>/reports/technical-graph.json from disk (or via the local HTTP API).

It is generic — scan whatever roots you pass; query from any consumer.

End-to-end flow

  features/*.yaml          repos (on disk)
         │                        │
         │                        ▼
         │              ┌─────────────────┐
         │              │  AST ingest     │  per repo: facts, CALLS, structural edges
         │              │  (CLI / local)  │
         │              └────────┬────────┘
         │                       │ merge-input.json per repo
         ▼                       ▼
  ┌──────────────────────────────────────────┐
  │  Merge (single scanId)                   │
  │  1. Dedupe + union repo graphs           │
  │  2. FE↔BE bridge (CALLS_REMOTE)          │
  │  3. Feature resolver (YAML + synthetic)  │
  │  4. technical-graph.json                 │
  └──────────────────┬───────────────────────┘
                     │
         ┌───────────┴───────────┐
         ▼                       ▼
   scans/<id>/ artifacts   AI stages (optional)
                           endpoint Q&A, feature
                           insights, summaries
         │
         ▼
   CLI / npm library / local API → agents

Query surface

Consumer	Entry
CLI	npx @igentify/helix-flowgraph query …
Library	queryFeatures, queryImpact, …
MCP	helix-flowgraph mcp --local
HTTP	HELIX_GRAPH_PATH=… npm run dev:api

See Ingest artifacts for on-disk layout.