OpenFS Roadmap

OpenFS Package Roadmap

A unified AI filesystem layer across every storage backend, runtime, and agent framework.

Core & Runtime

Done

@openfs/core

IFileSystem interface — read, write, ingest, ls, search. The contract all adapters implement.

Done

@openfs/wasm

sql.js SQLite WASM. Runs in browser, Bun, Node, Deno. Zero native deps. Primary dev adapter.

In Progress

@openfs/server

Hono HTTP API. Expose any adapter remotely. Add auth, streaming responses, WebSocket events.

Next

@openfs/cli

openfs ls, openfs cat, openfs grep, openfs push, openfs pull. Shell-native filesystem for AI.

Storage Adapters

Done

adapter-sqlite

Bun:sqlite + FTS5. Best local adapter. Production-ready.

Done

adapter-s3

S3/MinIO object storage. File-per-object, $contains search.

Done

adapter-chroma

Vector DB via Chroma. Semantic similarity search.

Done

adapter-postgres / mysql / turso

SQL adapters for team deployments. FULLTEXT search.

Next

adapter-redis

Redis HASH + RediSearch FTS. Sub-millisecond reads for hot knowledge.

Next

adapter-qdrant

Qdrant vector DB adapter. Better embedding support than Chroma for production.

Planned

adapter-r2

Cloudflare R2 + D1 adapter. Edge-native, globally distributed, free egress.

AI Agents

Done

@openfs/agent-wiki

Karpathy-style compounding knowledge base. Ingest sources, synthesize wiki pages, query with citations.

Done

@openfs/agent-wiki-mw

MediaWiki ↔ OpenFS bidirectional sync. Category tagging, synthesis map, Recent Changes integration.

Next

@openfs/agent-code

Code-aware agent. Indexes repositories, understands structure, answers questions about codebases.

Next

@openfs/agent-slack

Slack workspace agent. Ingests channels, threads, files. Answers questions about org knowledge.

Planned

@openfs/agent-notion

Notion workspace sync. Two-way: Notion pages ↔ OpenFS ↔ LLM synthesis.

Planned

@openfs/agent-confluence

Confluence ↔ OpenFS. Enterprise knowledge base integration.

Tooling

Done

@openfs/grep-optimizer

Two-phase grep optimization. Smart query planning before full scan.

Done

@openfs/indexer

Bulk document indexing pipeline. Chunking, dedup, metadata extraction.

In Progress

@openfs/playground

Astro live demo. Interactive adapter explorer. Embed in any page via iframe.

Next

VS Code Extension

Browse OpenFS in VS Code sidebar. Cat files, grep, ask questions about your codebase.

Next

OpenFS Shell

Interactive REPL: ls, cat, grep, ask, ingest. Like bash but for AI knowledge.

Planned

MCP Server

OpenFS as a Model Context Protocol server. Any Claude/LLM can read/write your FS natively.

This Wiki = OpenFS in Production

This MediaWiki instance is running on the full OpenFS stack: wasm (in-memory FS) → agent-wiki (LLM synthesis) → agent-wiki-mw (MW sync) → MediaWiki. Every page you read here was either written by Derek or synthesized by Claude Haiku from Derek's source pages. The OpenFS Terminal lets you grep, cat, ask, and manage the entire knowledge graph live.