Complete guides, API reference, and integration instructions for the open standard for AI agent memory.
Set up PLUR in Claude Code, Cursor, Windsurf, OpenClaw, or as a CLI tool. One command enables persistent memory across all your AI tools.
Read Guide →Learn the open standard for AI agent memory. Understand the data structure, activation model, and injection algorithm for portable memory.
Read Specification →Complete reference for all PLUR tools: plur_learn, plur_recall, plur_inject, plur_feedback, and more. Available via MCP, CLI, and multiple SDKs.
View API Docs →Retrieval recall, agent-task impact (Haiku + PLUR beats Opus without it), and operational metrics. 97.6% R@5 on LongMemEval-S.
View Benchmarks →How PLUR compares to other agent memory systems: Mem0, Letta, Mastra, Zep, and more. Detailed feature and architecture analysis.
See Comparisons →Articles on agent memory, the cost of context, learning architecture, and open standards for AI systems.
Read Articles →Run PLUR as an MCP server for Claude Code, Cursor, Windsurf, and any MCP-compatible agent. Full setup and troubleshooting guides.
MCP Documentation →Use PLUR via Python SDK, Hermes plugins, or command-line interface. Integrate with any AI framework or workflow.
View CLI Docs →Self-hosted deployment with SSO, SCIM, RBAC, audit logging, and team knowledge stores. Multi-machine sync via git.
Enterprise Docs →PLUR is listed as an official Memory Provider in the awesome-hermes-agent community directory. If you discovered PLUR there, here are your next steps:
If you're using Hermes and want persistent memory, install PLUR as a Memory Provider and it will auto-inject relevant engrams before each LLM call. No configuration needed — just one command to get started.
Questions? Issues? Feature requests?