Browse the library

Provides a centralized MCP server for managing multi-project memory banks, allowing AI agents to store and retrieve project-specific information remotely.

projectmem

riponcm

An open-source, local-first MCP server that provides memory and judgment for AI coding agents, recording issues, attempts, and decisions to prevent repeating past failures.

This tool provides a shared, searchable, local-first project memory system for AI coding agents, surviving sessions and agent switches.

Provides a shared, single-file local memory layer for AI agents like Claude and Cursor on Apple Silicon, offering sub-millisecond RAG over text, photo, and video.

A long-term memory system for AI coding agents like Claude Code and Cursor, allowing them to learn from corrections and retain knowledge across sessions locally.

OKF Agent Memory provides a Git-native, persistent project memory system for AI coding agents via an embedded MCP server, reducing token bloat.

Storybloq provides project memory, workflows, and context management for Claude Code and Codex agents, storing information in Git-trackable files.

Recall provides fully-local, durable project memory for Claude Code and Opencode, summarizing sessions without external models or API calls to reduce token waste.

Provides AI agents with persistent, collective, and semantic memory through a structured storage, retrieval, and graph visualization system.

Gives your agent a self-improving context engine that automatically documents its work, specs, and lessons in transparent Markdown files.

Provides always-on agent memory and workflow state management for Cursor, guiding the agent through development phases.

Provides persistent memory for Claude Code, allowing the agent to retain context and knowledge across sessions using an external database.

Provides a methodology and skills to build products with AI agents that remember, by turning documentation into a navigable knowledge graph.

RecallLoom provides persistent project memory for AI agents, saving context, decisions, and next steps in local project files without a database.

Provides agent skills for Codex OS and other agents, enabling persistent knowledge, verification, scheduled loops, and multi-agent orchestration.

Persistent local memory for Claude Code, Cursor and Codex over MCP or CLI that stores typed, sourced facts and retires old ones when they change.

This plugin helps Claude Code and Codex agents remember decisions, rules, and past approaches across sessions by saving them to a JEVMEM.md file in your repository.

A Claude Code plugin that hides large tool results the judge model deems irrelevant behind short stubs, restorable on demand, to save context tokens.

A Claude Code skill that opens replies with your name as a drift signal and keeps a TASK.md so a fresh session can resume where the last one stopped.

A Claude Code plugin with one MCP tool, search_ruvnet, that grounds Claude in real source for the RuvNet stack (RuVector, Ruflo, AgentDB and more).

Gives OpenCode, Claude Code and Codex persistent local memory that records decisions and dead ends and injects relevant context into later prompts.

This skill integrates AI agent's decisions and reasoning directly into Git history, creating a persistent log of development choices and context across sessions.

This plugin provides always-current official Claude API and platform documentation directly within Claude Code, with instant search and interactive courses.

Provides a Model Context Protocol (MCP) server that learns from your development patterns to give AI coding agents continuous, project-centric context awareness.