Agent frameworks & infrastructure
Agent memory & context
4,032 agents · most starred first
Agent memory & context, page 35 of 41
Universal agent memory plugin (ForgetForge): recall-centric retention, tiered forgetting, Rust scoring engine for Hermes, Claude Code, and Codex.
Haystack integration for Hindsight - persistent memory for AI agents
Superagent safety integration for Hindsight - guard and redact memory operations
Cache-optimized LLM conversation history management with static/dynamic system prompts, transition modes, and compaction hooks.
Three-tier memory with trust-aware context ledger for multi-agent LLM systems
Get Ready With Me - The universal AI agent handoff tool. Never explain your codebase to a new agent again.
Local-first agent memory: an Obsidian markdown vault as source of truth, with a rebuildable DuckDB index.
Tenant isolation and per-tenant usage metering for LangGraph checkpointers and stores
TypeScript & Node.js SDK for HipCortex - the persistent causal memory & token optimization substrate for autonomous AI agents.
Mission working memory for LLM agents: session graph + pin_map, not a RAG corpus
Pull requests for AI memory.
MCP checkpoint server: atomic state coordination for AI agents, cron workers, and multi-agent systems
Microsoft Agent Framework integration for Hindsight - persistent memory for AI agents
Context garbage collector for long-running LLM agents — offload and recall to save tokens (Claude Code & Codex via MCP)
Shared memory and tool layer for Perseus agents — MemoryEntry, Elastic/Engram backends, decision log, knowledge graph, project context.
TypeScript SDK for AgentState — conversation history for AI agents
Persistent memory for Claude Code — checkpoint-based replanning across sessions
LlamaIndex memory integration for Recollect
SuperMemory: MCP-first learning memory layer for Claude, Cursor, and agent workflows
SuperMemory: MCP-first learning memory layer for Claude, Cursor, and agent workflows
Fractal memory system for OpenCode with semantic search and automatic compression.
Halo host adapter for LangChain agents and LangGraph graphs: a wrapToolCall encode middleware (results become a shape map, out of context) plus a single halo_fetch navigation tool. installHalo() wires it in one call.
Framework-agnostic agent memory system — Postgres + pgvector + MCP
LangChain and LangGraph integration for the Mnemosyne memory substrate. BaseChatMessageHistory, BaseMemory, and BaseRetriever adapters with hybrid BM25+vector recall, per-session thread tracking, automatic redaction, TTL support.
JSONL-backed persistence for AI agent insights
Automatic token compression for AI agent tool outputs. 60-90% fewer tokens, full explainability.
Local-first persistent memory for AI agents over MCP. Ground truths, negative learning (recorded failures block repeats), decision causality, hybrid recall, live memory streams, run-attributed fleet memory — a structured SQLite memory on your machine, no
Compress tool outputs, logs, JSON and chat history before each LLM call. Fewer input tokens, same answers. Adapters for OpenAI, Anthropic, Gemini and the Vercel AI SDK. Runs against the local Headroom proxy (pip install "headroom-ai[proxy]").
LangChain integration for the Unison Brain — memory, chat history, and retriever
LlamaIndex memory provider backed by the Unison brain — long-term agent memory via a knowledge graph.
Context compression and planning for RAG and AI agents: fit retrieved documents, chat history and tool output into a token budget.
Mimir persistent memory tool for CrewAI agents
Mimir persistent memory backend for LangGraph agents
AutoGen v0.4 MemoryStore adapter for the Mnemosyne memory substrate. Implements the MemoryStore protocol with per-agent container scoping, hybrid BM25+vector recall, async support for high-throughput multi-agent runs.
CrewAI long-term memory adapter for the Mnemosyne memory substrate. Implements the LongTermMemory protocol via duck typing, with per-agent and per-task scoping, hybrid BM25+vector recall, idempotent writes.
OpenAI Agents SDK lifecycle hooks for the Mnemosyne memory substrate. on_turn_start, on_turn_end, search, and store hooks backed by the substrate, container-scoped, with automatic redaction.
Self-Learning Index Memory for AI agents
MCP-native memory server with built-in governance for AI agents
TypeScript SDK for RetainDB - the memory bus for agentic systems
Dory Agentic SDK: LFS context management with AI SDK tool loops for long-running agents
<p align="center"> <img src="https://raw.githubusercontent.com/cacheplane/dawnai/main/docs/brand/dawn-logo-horizontal-black-on-white.png" alt="Dawn" width="180" /> </p>
Never lose work when an agent fails.
Shared operational memory and coordination for mixed AI agent fleets.
Shrink LLM context windows by removing noise, redundancy, and long-tail detail — without losing signal.
Client library for the opengater-storage service
Automatic memory capture and recall for AI agent workflows
Local SQLite memory engine for omp agents
LangGraph checkpointer adapter for MemForks — anchors agent state on Sui with branch-aware memory.
Agent Memory HTTP SDK and lightweight Agent adapter
Pieces is a private, on-device, long-term memory layer for macOS, Linux, and Windows that proactively captures your work context as it happens. Over time, Pieces becomes your very own “internet of all context”—browse your artificial memories in the Pieces Desktop app, or plug them into the agents you already use: OpenClaw, Hermes, Claude, Cursor, Codex, etc.
Memlink — MCP server for persistent AI agent memory. Self-hosted, one URL, no cloud.
Repository-native agent memory, skills, AGENTS.md, llms.txt, and knowledge workflows for coding agents.
JSON Schemas and Python contracts for AI agent memory integrity: evidence, candidate, ledger, taste, state, and ContextPack planes.
A virtual shell-over-database filesystem for AI agents: a strict, append-only memory layer with a frozen shell dialect.
Persistent memory for AI agents — store, retrieve, and recall context across sessions.
Collective knowledge store for AI agents — publish facts at runtime, query shared learning
AgentRelay Secure Memory Persistence and Recall integrations for AI frameworks.
Explainable human-like memory layer for AI agents.
MemGPT memory architecture as a Python sidecar for the openclaw-memgpt plugin. Includes Anthropic/Claude compatibility patches against the original MemGPT (Apache 2.0).
Git for AI agent memory — TypeScript SDK for MemForks on Sui
OpenViking memory adapters for LangChain.js and LangGraph.js — retriever, store, agent tools, chat history, and context middleware.
Multi-tiered memory and knowledge base plugin for TUI coding assistants
Agent memory in a single SQLite file — no vector DB, no server, no cloud. Keyword recall on stdlib alone; torch-free hybrid semantic search via model2vec + sqlite-vec. Python lib + MCP server + CLI.
Single-file embedded database for AI agents. SQL + Vector Search + Full-Text Search + Hybrid Queries + Memory Graphs.
O(1) hyperbolic memory for any LLM — 260 bytes per session forever
Standalone AI agent memory system — SQLite, FTS5, embeddings, 4-layer recall, sleep maintenance
Pi extension that keeps noisy tool output out of your context window. Automatically rewrites shell commands through Hypa for local, deterministic compression, context-aware file tools, and recoverable evidence.
TypeScript SDK for LocusGraph memory system with LangChain integration
Semantic-causal CRDT for agent-mutable world state
Reusable lessons/memory engine for Claude Code: frontmatter lessons, auto-CATALOG, offline retrieval, path-triggered lessons, auto-maintenance, parallel-session and sub-agent guards.
OKF-aware virtual filesystem backend for LangChain Deep Agents (Open Knowledge Format).
Compression with a quality contract — cache-aware, causally-pruned context compression for agentic runtimes, gated by a statistical non-inferiority test.
Give your LLM agent a mood — Yerkes-Dodson stress modeling, Russell affect mapping, allostatic regulation, and a self-tuning 3-arm bandit. Sub-µs core ops, every op <1ms, 0 tokens, fully serializable.
LangChain + LangGraph integration for Anansi — retriever, memory tools, and graph-state memory annotation
Framework-agnostic memory tools for Anansi — remember/recall tool definitions for the Vercel AI SDK, OpenAI function calling, and any agent framework
Agentic memory for Pi — verbatim session capture + grep retrieval
Pulse portable memory for AI agents: MCP server (pulse mcp) + installer/CLI in one package.
ACA-governed checkpointer for LangGraph.js — adds source_tier, provenance, namespace isolation, and audit trail to LangGraph state
A zero-dependency skill memory, testing, and refinement engine for AI agents.
OmniAgentMemory — a namespaced, structured memory service for Claude Code and other agents
LangChain.js BaseChatMemory implementation backed by Mnemosyne. Persistent, searchable, container-scoped chat history with hybrid BM25+vector recall. Replaces LangChain's in-process BufferMemory and ZepMemory. Per-conversation thread tracking, automatic r
Honcho memory for the Vercel AI SDK -- persistent identity, theory-of-mind, dialectic reasoning
Library with a Tigris object-storage implementation of LangGraph checkpoint saver, with zero-copy bucket-fork branching.
Give your OpenClaw agent lasting memory: structured facts, semantic search, auto-capture & recall, decay, optional credential vault. Part of Hybrid Memory v3.
Official LangChain.js and LangGraph.js integration for HyperspaceDB
Upstash AgentKit adapter for the Vercel Eve agent framework: memory tools, an Upstash Redis memory backend and provider for eve memory slots, Redis-Search tools, a rate-limit gate, an Upstash Box sandbox backend, and cached tools.
A-MEM agentic memory backend for OpenClaw — TypeScript rewrite
Agentic systems specialist agent for Bluetemberg — agent memory design, state management, orchestration patterns.
Persistent memory infrastructure for AI agents - Memory that adapts. Intelligence that compounds.
Embedded, file-backed GraphStore for AgentForge (Kùzu)
Forge persistent memories for your AI agent. 9 MCP tools + 5 auto-engines. Powered by Shelby decentralized cloud.
Tiered, token-budget-aware memory for any LLM agent
OpenCode plugin for LLM prompt caching, token savings, and KV cache reuse across AI agent sessions
Embedded memory for TypeScript agents. Local-first by default. Pluggable to edge and serverless. Extract facts, embed them, and recall scoped memories from your app process, with pluggable vector storage and framework adapters.
Local-first coding agent memory for Claude Code and OpenAI Codex
Git for Agent Memory: Copy-On-Write vector branching for embedded multi-agent memory. Branch a base memory in ~0.5ms / 162 bytes regardless of base size — 83x faster, 3000x smaller than full-copy snapshots. Exact read-through queries (parent ∪ edits, chil
Passive Obsidian vault extension for pi. Watches @agent markers, dispatches forked subagents (Miner, Broadcaster, Heavy-Lifter), stores in LanceDB with vector + FTS + graph. Multi-agent 'Drop & Forget' workflow for the pi agent ecosystem.
Local-first Markdown-backed MCP memory and automatic project-context engine for multi-agent coding, indexed with Postgres and pgvector.
Cross-session memory layer for the Vercel eve agent framework
Eidetic — Agentic Memory MCP Server