PLUR vs Mastra

Choose Mastra for memory built into its TypeScript agent framework. Choose PLUR when memory must outlive and cross frameworks.

Bottom line: Mastra if you're building inside its TS framework and want memory integrated there; PLUR if memory must be framework-agnostic — an open format shared across every tool over MCP.

What each is best at

• Mastra — a TypeScript agent framework with observational memory built in. Its strength is developer experience within the framework: memory, workflows, and tools in one TS stack.
• PLUR — a memory layer independent of any framework: a typed, open YAML engram format (published spec) that runs fully local and is shared across tools over MCP.

Where they differ (that matters)

• Framework-bound vs framework-agnostic. Mastra's memory lives inside Mastra (TS). PLUR's memory is separate and portable — it spans Claude Code, Cursor, Windsurf, OpenClaw, Hermes, and Mastra itself.
• Open portable format. PLUR's memory is an open engram format you own, read, and move; Mastra's memory is a framework feature, not an interchange spec.
• Local-first, sovereign. PLUR's default is fully local with no phone-home; erasure is a real file delete.

Recall — table stakes, not the deciding factor

Both are competitive; quality is table stakes. PLUR reaches 97.6% R@5 on LongMemEval-S, fully local. (Mastra's memory numbers are blog-level; we don't run a head-to-head.) The decision is portability, not a recall delta. (LongMemEval-S · n=500 · chunk · canonical-doc; R@5 = evidence in the top-5, not answer accuracy; measured on our own plur-bench harness, public with our paper.)

Choose Mastra if

you're building your agents in Mastra's TypeScript framework and want memory tightly integrated there.

Choose PLUR if

your memory must be framework-agnostic and portable — shared across multiple tools, owned as an open format, running locally.

Install PLUR

npx @plur-ai/mcp init      # Claude Code / Cursor / Windsurf (any MCP client)
openclaw plugins install @plur-ai/claw && openclaw gateway --force   # OpenClaw
pip install plur-hermes    # Hermes Agent (Python)

Engrams are stored locally as files under ~/.plur/. Connect over MCP from Claude Code, Cursor, Windsurf, OpenClaw, or Hermes — including alongside Mastra.

FAQ

Mastra vs PLUR — which should I use? Mastra for memory inside its TS framework; PLUR for a portable, open-format memory that crosses frameworks.

Can I use PLUR with Mastra? Yes — PLUR is a framework-agnostic memory layer over MCP; it complements rather than replaces a framework.