The content-ingestion pipeline gives contributors a route through extraction, tagging, embeddings, and search.
Documents
└──
The file, explained
What makes it useful
Supermemory's CLAUDE.md is primarily an architecture and command reference. It describes the web and MCP applications, API routes, shared dependencies, and the content-processing workflow. It also lists deployment bindings and conventions for generated schema types. The guide has gaps, including an unfinished Web Application section, so its value is orientation rather than a complete change or verification policy.
Quoted passages are verbatim. Open one to see it in the source.
01 / Architecture as narrative
Describe the content-processing path
The named workflow connects content detection and extraction to summarization, embeddings, chunking, and space relationships. That sequence helps locate where a change in ingestion behavior belongs.
Source excerpt starting at line 70.
70All content goes through the `IngestContentWorkflow` which handles:71- Content type detection and extraction72- AI-powered summarization and automatic tagging73- Vector embedding generation using Cloudflare AI74- Chunking for semantic search optimization75- Space relationship management
02 / Verification by change type
Distinguish workspace and application commands
Root commands operate across the monorepo, while the web application has its own development and build commands. The file supplies both scopes rather than leaving working-directory assumptions implicit.
Source excerpt starting at line 15.
15### Root Level (Monorepo)16- `bun run dev` - Start all applications in development mode17- `bun run build` - Build all applications18- `bun run check-types` - Run TypeScript checks across all apps19- `bun run format-lint` - Format and lint code using Biome2021### Web Application (`apps/web/`)22- `bun run dev` - Start Next.js development server23- `bun run build` - Build Next.js application24- `bun run lint` - Run Next.js linting
03 / Architecture as narrative
List the deployment dependencies
The environment section names the Cloudflare bindings and the connection-import schedule. These details explain which external capabilities the processing architecture expects.
Source excerpt starting at line 77.
77### Environment Configuration78- Uses `wrangler.jsonc` for Cloudflare Workers configuration79- Supports staging and production environments80- Requires Cloudflare bindings: Hyperdrive (DB), AI, KV storage, Workflows81- Cron triggers every 4 hours for connection imports
Put it to work
Borrow this for your repo
01Describe the processing stages a contributor must trace when changing behavior.
02Label commands by workspace or application scope.
03List required runtime bindings beside the workflows that depend on them.
Supermemory's file tells an agent how the codebase works. It cannot tell it which bug three customers hit this week. Modem keeps that context current and attaches it to the work.
