# Wolbarg

> Shared semantic memory for multi-agent systems. Wolbarg Workspace is the product for Cursor, Claude Code, and Codex. The wolbarg SDK powers custom agents.

## Prefer these URLs

- [Wolbarg Workspace](https://wolbarg.com/docs/workspace): Shared project truth for coding agents (`@wolbarg/workspace`)
- [Quick Start](https://wolbarg.com/docs/quick-start): remember() and recall() in under a minute
- [Concepts](https://wolbarg.com/docs/getting-started): What Wolbarg is and why it exists
- [Connectors](https://wolbarg.com/connector): Workspace coding hosts + AI SDK adapters
- [Architecture](https://wolbarg.com/docs/architecture): Subsystems from Application → Database
- [Benchmarks](https://wolbarg.com/benchmarks): Interactive performance numbers
- [llms-full.txt](https://wolbarg.com/llms-full.txt): Full docs as one Markdown file for AI ingestion
- [GitHub SDK](https://github.com/wolbarg/wolbarg): SDK source repository
- [GitHub Workspace](https://github.com/wolbarg/workspace): Workspace product repository

```bash
npx @wolbarg/workspace init
```

## Features

- [Workspace](https://wolbarg.com/docs/workspace): Coding-agent product on npm
- [Connectors](https://wolbarg.com/connector): Workspace hosts + AI SDK adapters
- [Hybrid Search](https://wolbarg.com/docs/hybrid-search): Semantic vectors fused with BM25
- [Storage Providers](https://wolbarg.com/docs/storage/sqlite): Pluggable backends
- [Provider Architecture](https://wolbarg.com/docs/providers): Embeddings, LLM, OCR, vision, rerankers
- [Document Ingestion](https://wolbarg.com/docs/document-ingestion): PDF, DOCX, Markdown, images
- [Compression Pipeline](https://wolbarg.com/docs/compression): LLM summarization of memories
- [Metadata Filtering](https://wolbarg.com/docs/metadata-filtering): eq, contains, AND/OR/NOT filters

## Documentation index

# Documentation


- **Getting Started**
- [Quick Start](/docs/quick-start): Construct Wolbarg, remember facts, recall them, and optionally use hybrid search and document ingest.
- [Installation](/docs/installation): Install Wolbarg, optional peers, and a clean project layout that keeps providers swappable without refactoring agent code.
- [Concepts](/docs/getting-started): What Wolbarg is, why it exists, and the provider philosophy behind its architecture.
- [Wolbarg Workspace](/docs/workspace): Shared, reconciled project truth for Cursor, Claude Code, and Codex — install from npm with @wolbarg/workspace.
- [Observability & Studio](/docs/observability): Enable Wolbarg telemetry and explore it in Wolbarg Studio — live dashboard, Trace Explorer waterfalls, and ops surfaces.

- **Connectors**
- [Connectors](/docs/connectors): Docs index for Wolbarg connectors — the product listing lives at /connector.

- **AI SDKs**
- [Overview](/docs/integrations): Connect Wolbarg to agent frameworks and AI SDKs without putting framework types in the core package.
- [Vercel AI SDK](/docs/integrations/vercel-ai): Official @wolbarg/vercel-ai middleware — automatic recall, system injection, and remember with generateText and streamText (AI SDK v7+).
- [OpenAI Agents SDK](/docs/integrations/openai): Official @wolbarg/openai Session — persist AgentInputItem history, semantic remember, and recall injection via createWolbargSessionInputCallback.
- [LangChain / LangGraph](/docs/integrations/langchain): Official @wolbarg/langchain — WolbargMemory (BaseMemory), WolbargStore (LangGraph BaseStore), and createWolbargTools.
- [LlamaIndexTS](/docs/integrations/llamaindex): Official @wolbarg/llamaindex — wolbargBlock / WolbargMemoryBlock for createMemory({ memoryBlocks }).
- [Mastra](/docs/integrations/mastra): Official @wolbarg/mastra Processor — createWolbargProcessor for shared semantic recall and remember (not a Storage rewrite).

- **Configuration**
- [Configuration](/docs/configuration): Required and optional constructor options for Wolbarg — organization, storage, embedding, providers, concurrency, embeddingCache, memory.dedupe, and AbortSignal.
- [Provider Architecture](/docs/providers): Embedding, LLM, keyword search, reranker, OCR, vision, and chunking providers in Wolbarg.

- **Storage**
- [SQLite Backend](/docs/storage/sqlite): Local-first SQLite storage with WAL, sqlite-vec vectors, and FTS5 keyword indexing.
- [PostgreSQL Backend](/docs/storage/postgresql): Shared PostgreSQL storage with connection pooling, JSONB metadata, and optional pgvector.

- **Search & Retrieval**
- [Semantic Search](/docs/search): How Wolbarg embeds queries and retrieves memories by cosine similarity.
- [Hybrid Search](/docs/hybrid-search): Combine semantic vectors with BM25 keyword scores for more robust recall. Fail-closed since 0.6.0.
- [Rerankers](/docs/rerankers): Optional cross-encoder reranking and MMR diversification for recall results. Fail-closed since 0.6.0 — throws RerankError on provider failure.
- [Metadata Filtering](/docs/metadata-filtering): Filter recall with meta.eq, contains, comparisons, and AND/OR/NOT boolean trees.
- [Compression Pipeline](/docs/compression): Summarize and archive memories with an optional LLM via compress().

- **Ingestion**
- [Document Ingestion](/docs/document-ingestion): Parse PDF, DOCX, Markdown, and other documents into chunked semantic memories with ingest().
- [Image Ingestion](/docs/image-ingestion): Store image-derived text as semantic memory using OCR and vision providers.
- [Chunking](/docs/chunking): Pluggable chunking strategies for document ingest — fixed, sentence, paragraph, markdown, heading.
- [OCR](/docs/ocr): Extract text from images with tesseract.js during Wolbarg ingest.
- [Vision Models](/docs/vision): Caption and describe images with Gemini or OpenAI vision providers during ingest.

- **Architecture**
- [Architecture](/docs/architecture): How Wolbarg subsystems connect — Application, Wolbarg, Storage, Retrieval, Providers, and Database.
- [Performance](/docs/performance): Tuning guidance for recall latency, ingest throughput, and storage growth in Wolbarg.
- [Concurrency](/docs/concurrency): Multi-process SQLite write safety in Wolbarg 0.4 — BEGIN IMMEDIATE, busy_timeout, exponential backoff, WOLBARG_STORAGE_LOCKED, and published multi-writer benchmarks.
- [Real-time events](/docs/realtime-events): Subscribe to memory changes with wolbarg.subscribe() — in-process EventEmitter for SQLite, LISTEN/NOTIFY for Postgres, with filters and safe callbacks.
- [Embedding cache](/docs/embedding-cache): Transparent hash(content)+model embedding cache in Wolbarg 0.4 — cut provider cost and latency on repeated text with optional LRU and TTL.
- [Memory upsert & deduplication](/docs/memory-upsert): Opt-in write-time exact and near-duplicate detection in Wolbarg 0.4 — update existing facts instead of inserting duplicates, with RememberResult.action and update().
- [Graph memory](/docs/graph-memory): Graph memory was removed in Wolbarg 0.6.0. Use metadata or an external store for relationships.
- [Benchmarks](/docs/benchmarks): Methodology and published results for Wolbarg v0.4.0 — dual-backend v4 stress, embedding cache, multi-process concurrency, and how to reproduce locally.

- **API**
- [API Overview](/docs/api): Index of the Wolbarg public API — constructor, remember, rememberFromMessages, update, recall, ingest, forget, history, subscribe, AbortSignal, and lifecycle.
- [Wolbarg](/docs/api/wolbarg): Lifecycle methods for the Wolbarg class — constructor options including concurrency, embeddingCache, memory.dedupe, ready, close, and subscribe.
- [remember()](/docs/api/remember): Store a semantic memory with embedding, optional metadata, optional dedupe, and RememberResult.action.
- [rememberFromMessages()](/docs/api/remember-from-messages): Experimental conversation → memory bridge — store chat turns as memories (raw or LLM extract).
- [update()](/docs/api/update): Explicitly update an existing memory by id — content and/or metadata — returning RememberResult.
- [recall()](/docs/api/recall): Semantic and hybrid search with filters, thresholds, MMR, and rerank. Fail-closed hybrid/rerank since 0.6.0. Supports AbortSignal.
- [ingest()](/docs/api/ingest): Parse documents into chunked semantic memories.
- [forget()](/docs/api/forget): Archive or delete memories by id or metadata/agent filter. Supports AbortSignal.
- [history()](/docs/api/history): Read audit events for remember, forget, compress, and related operations.
- [subscribe()](/docs/api/subscribe): Register real-time callbacks for memory change events — filter by organization, agent, and event type.
- [stats() / clear()](/docs/api/lifecycle): Introspection and organization-scoped wipe helpers.

- **Generated Reference**
- [API Reference](/docs/api/reference): Export catalog and curated method pages for the wolbarg package.
- [Generated Exports](/docs/api/reference/generated): Compact auto-generated catalog of public exports from the Wolbarg package entrypoint.

- **Examples**
- [Examples](/docs/examples): Independent Wolbarg examples — basic memory, conversation bridge, Vercel AI adapter, hybrid search, storage backends, ingest, and providers.
- [Example — Basic Memory](/docs/examples/basic-memory): Minimal remember and recall with SQLite and OpenAI embeddings.
- [Example — Conversation memory](/docs/examples/conversation-memory): Chat transcript → rememberFromMessages → recall without hand-rolled extraction.
- [Example — Vercel AI memory](/docs/examples/vercel-ai-memory): Use @wolbarg/vercel-ai middleware with wrapLanguageModel for automatic recall and remember (AI SDK v7+).
- [Example — Hybrid Search](/docs/examples/hybrid-search): Enable BM25 keyword search and fuse it with semantic recall.
- [Example — PostgreSQL](/docs/examples/postgresql): Shared PostgreSQL storage with the pg peer dependency.
- [Example — SQLite](/docs/examples/sqlite): File-backed and in-memory SQLite storage examples.
- [Example — Image Memory](/docs/examples/image-memory): Ingest a PNG with OCR and vision enrichment.
- [Example — PDF Memory](/docs/examples/pdf-memory): Ingest a text-layer PDF into chunked memories.
- [Example — Metadata Filtering](/docs/examples/metadata-filtering): Scope recall with meta helpers and agent filters.
- [Example — Compression](/docs/examples/compression): Summarize memories with an LLM via compress().
- [Example — Providers](/docs/examples/providers): Mix OpenAI embeddings, Ollama LLM, and BM25 keyword search.
- [Example — Rerankers](/docs/examples/rerankers): Attach a Jina reranker and enable rerank on recall.
- [Example — OCR](/docs/examples/ocr): Extract text from images with tesseract during ingest.

- **Guides**
- [Multi-Agent Memory](/docs/guides/shared-memory): Share one Wolbarg instance across concurrent agents with agent-scoped filters.
- [What's New](/docs/guides/whats-new): Wolbarg 0.6.0 release notes — fail-closed hybrid/rerank, Postgres SSL and schema, AbortSignal, concurrency hardening, and graph memory removal.
- [Production](/docs/guides/production): Operator guidance for deploying Wolbarg 0.6 — SQLite vs Postgres, SSL, schema, pooling, fail-closed hybrid/rerank, backups, and troubleshooting.
- [Limitations](/docs/guides/limitations): Honest boundaries of Wolbarg 0.6 — peers, PDF quality, SQLite, Postgres SSL/pool, fail-closed hybrid/rerank, and removed graph memory.
- [Best Practices](/docs/guides/best-practices): Practical guidance for production Wolbarg usage — scoping, filters, peers, lifecycle, and provider isolation.

- **Reference**
- [Types](/docs/reference/types): Key public TypeScript types in Wolbarg.
- [Errors](/docs/reference/errors): Typed error hierarchy for Wolbarg 0.6 — ValidationError, RerankError, StorageLockedError, CancellationError, and more.
- [init() Compatibility](/docs/reference/init-compat): v0.1 init() API remains supported as a shim.
- [FAQ](/docs/faq): Frequently asked questions about Wolbarg installation, providers, storage, and retrieval.
- [Migration](/docs/migration): Upgrade Wolbarg — 0.5 → 0.6 production hardening, earlier additive upgrades, and AgentOrc → Wolbarg rebrand notes.

## Optional

- [Sitemap](https://wolbarg.com/sitemap.xml): Full crawl map
- [FAQ](https://wolbarg.com/docs/faq): Common questions
- [Migration](https://wolbarg.com/docs/migration): Upgrade from 0.1 to 0.2