iiRecord
Agentic AI Atlas · SMFS (Semantic Memory File System) Overview
page:docs-supermemory-research-raw-02-smfs-overviewa5c.ai
II.
Page JSON

page:docs-supermemory-research-raw-02-smfs-overview

Structured · live

SMFS (Semantic Memory File System) Overview json

Inspect the normalized record payload exactly as the atlas UI reads it.

File · wiki/docs/supermemory-research/raw/02-smfs-overview.mdCluster · wiki
Record JSON
{
  "id": "page:docs-supermemory-research-raw-02-smfs-overview",
  "_kind": "Page",
  "_file": "wiki/docs/supermemory-research/raw/02-smfs-overview.md",
  "_cluster": "wiki",
  "attributes": {
    "nodeKind": "Page",
    "sourcePath": "docs/supermemory-research/raw/02-smfs-overview.md",
    "sourceKind": "repo-docs",
    "title": "SMFS (Semantic Memory File System) Overview",
    "displayName": "SMFS (Semantic Memory File System) Overview",
    "slug": "docs/supermemory-research/raw/02-smfs-overview",
    "articlePath": "wiki/docs/supermemory-research/raw/02-smfs-overview.md",
    "article": "\n# SMFS (Semantic Memory File System) Overview\n\nSource: https://supermemory.ai/docs/smfs/overview\n\n## Core Concept\n\nSMFS enables agents to access Supermemory containers using standard filesystem operations. \"Memory your agent can grep.\" The system provides real directory mounting where agents can use familiar shell commands like `ls`, `cat`, and `grep` without requiring SDK integration or embedding knowledge.\n\n## Key Features\n\n### Semantic Search\n\nThe grep functionality operates semantically by default, surfacing relevant content across the entire container ranked by relevance. Users can append flags to revert to exact-match searching.\n\n### Context Optimization\n\nMemory paths get distilled and indexed by Supermemory, preventing them from consuming model context tokens. A virtual `profile.md` file provides a live summary of container contents.\n\n### Synchronization\n\nBackground bidirectional syncing ensures local reads hit cached data while writes push updates to Supermemory.\n\n## Deployment Options\n\n### 1. Mount Binary\n\nFor environments with real filesystems (Claude Code, Cursor, Docker). Uses NFSv3 on macOS and FUSE on Linux.\n\n### 2. Bash Tool\n\nFor serverless/edge deployments (Lambda, Cloudflare Workers, Vercel). Available as TypeScript (`@supermemory/bash`) and Python (`supermemory-bash`) packages.\n\n## Supported Platforms\n\nIntegration guides exist for Daytona, E2B, Vercel AI SDK, and Cloudflare Workers.\n",
    "documents": []
  },
  "outgoingEdges": [],
  "incomingEdges": [
    {
      "from": "page:docs-supermemory-research",
      "to": "page:docs-supermemory-research-raw-02-smfs-overview",
      "kind": "contains_page"
    }
  ]
}