engram vs agentfs

SQLite-backed state and memory for AI agents compared

engram and agentfs both use SQLite to make useful agent data persist beyond a single run. engram focuses on searchable project knowledge for coding agents, while agentfs stores files, key-value state, and tool-call history as an inspectable filesystem database.

engramagentfs
LanguageGoRust
LicenseMIT—
Stars7k3.5k
Forks724204
Last analyzedOct 3, 2026Oct 13, 2026

Key differences

  • engram is a Go-based memory system for preserving project decisions, discoveries, and handoffs; agentfs is a Rust-based storage layer for files, key-value state, and tool-call records.
  • engram offers MCP, CLI, HTTP API, and terminal UI access; agentfs provides SDKs and a CLI, with filesystem mounts through FUSE on Linux and NFS on macOS.
  • engram emphasizes project-aware memory retrieval and curation; agentfs emphasizes portable, queryable state that can be inspected or restored from a database file.
  • engram lists an MIT license; agentfs is identified as MIT licensed in its README.
  • agentfs describes itself as beta and advises caution with production data; engram’s stated considerations focus on memory curation and installation or upgrade paths.

Choose engram if you…

  • need persistent, searchable project knowledge for coding agents across sessions.
  • want to connect through MCP or use a CLI, HTTP API, or terminal UI.
  • prefer local SQLite memory with optional Git Sync or Cloud replication for sharing.
Read the engram analysis →

Choose agentfs if you…

  • need files, key-value state, and tool-call history stored together.
  • want to inspect or restore agent state by copying a SQLite database file.
  • need SDK access in TypeScript, Python, or Rust, or filesystem mounts on supported platforms.
Read the agentfs analysis →

Compare the numbers side by side, or add more projects

This comparison is generated with AI from the OSRepos analyses of both projects. Always check each project's repository and documentation before choosing.