Agent frameworks & infrastructure
Agent memory & context
4,032 agents · most starred first
Agent memory & context, page 18 of 41
GenPark AI Agent Skill - Belief revision logic engine (AGM postulates) detecting factual and logical contradictions, pruning obsolete beliefs, and preserving consistent world states.
GenPark AI Agent Skill - Computes Personalized PageRank (PPR) over agent memory graphs for context-sensitive entity relevance and associative recall.
Second brain for AI agents: a plain-markdown Obsidian vault an agent can read, write to and navigate. Template, filing conventions and the memory model. For research, startups and side projects.
Memory for AI agents with source tracking, corrections, and history. Python and TypeScript SDKs.
Local memory + semantic search for AI agents. Indexes code and documents on your machine, queryable via MCP. Private, offline, fast. macOS arm64 + Linux x86_64.
InterPlanetary Artificial Intelligence (IPAI): store your personal AI model leveraging on InterPlanetary Data Wallet (IPDW). It enables to build a real censorship resistant research swarm without centralised repositories. Browser based inferences, Personal fine-tuning, Zero-backend privacy, and much more...
Agent memory for any AI agent: full conversations kept as plain, related tables any tool, model or person can read and share — Claude Code first
Knowledge graph memory for Claude Code and Codex CLI — persistent nodes, typed edges, recall at the moment it applies, self-maintaining. One local memory both agents share.
A living knowledge distillation system that self-evolves by reading technical content. Self-improving skill/knowledge pipeline for AI agents.
Okto Neuron: local-first knowledge graph memory for agents
Local, private, and portable lifelong memory management for AI agents.
Sleep-cycle memory management for coding agents — consolidate offline, verify against source, forget on schedule
A lightweight dynamic memory engine for LLM Agents. Core: Decay weights + Spatiotemporal Graph + Explainable Retrieval. No Docker, No Vector DB, Just SQLite.
Secure, persistent, and cross-session agentic memory for commerce state. This skill enables OpenClaw nodes to maintain granular historical context of vendor negotiations, discount eligibility, and user preferences over long-term fulfillment cycles.
Gaste menos cota, nunca perca contexto e tenha um diário de tudo que a IA fez. Governança, memória (ai-memory) e nota viva no Obsidian para Claude Code, Codex e Antigravity.
LogMem server: shared memory for agent teams (implements ACMI).
Give one OpenClaw session durable memory, topic-aware continuity, and bounded token growth.
Universal AI Agent Long-Term Memory Engine — Hybrid Vector+BM25, cross-encoder reranking, Weibull decay, LLM smart extraction, semantic dedup, tiered content. Zero-ops LanceDB backend.
A dual-mode tool providing long-term memory management for Claude Code and other MCP clients. Features both a CLI interface for direct commands and an MCP server for programmatic access.
🧠 Teach your AI agents once. Reuse everywhere. A Git-native registry for AI skills and team knowledge.
Nostr-native agent memory system
Self-hosted MCP memory server — 35 tools for persistent context across AI coding sessions. Runs on Docker with PostgreSQL + pgvector + Ollama.
Tripartite memory SDK for AI agents: semantic recall, structural graph awareness, and relational state — before agents act.
Give your assistant perfect memory, and still own the data.
Resk-Caching is a Bun-based backend library and server designed for secure caching, embeddings orchestration, and vector database access. It prioritizes security, high performance, and deep observability.
With ME_GPT, you can store your cherished memories in a local vector database and leverage them to give Chat GPT the extraordinary ability to clone your unique personality and knowledge.
AI agent backend with long-term memory, document processing, semantic search, and dynamic function orchestration.
HIEF (Hybrid Intent-Evaluation Framework) Persistent Memory Layer for AI Coding Agents
CogmemAi — Cognitive Memory for Any Ai System. Autonomous robots, self-driving vehicles, defense systems, coding assistants, and more. 91% LoCoMo benchmark — above human performance.
A protocol file for your AI agent. Drop a SMITH.md in any directory and any coding agent becomes a long-term collaborator that remembers who you are.
🧠 Open-source AI reasoning research platform — persistent reasoning graphs, multi-agent swarms, Graph-of-Thoughts, and MemGPT-style memory. Built with Claude Opus 4.6 by Anthropic. Explore, verify, and fork AI thinking in real-time.
Stop re-teaching your AI. Turn past Claude Code sessions into persistent memory.
Persistent structured memory for Claude Code. No infrastructure required — just markdown files and git.
The open-source benchmark for LLM memory decay. Measure how Naive, RAG, Chunked RAG, Cascading, and SummaryMemory degrade over 100 conversation turns. Ebbinghaus forgetting curves, 5-provider LLM eval, multi-seed CI. No API key needed.
Give your AI a perfect, infinite memory. A local-first, zero-LLM memory system and Model Context Protocol (MCP) server designed to give AI assistants (like Claude, ChatGPT, and custom agents) a searchable, structured "Memory Palace."
High-recall conversational memory retrieval. 93.9% LoCoMo, 98.4% R@5 on LongMemEval — no LLM required. Local-first, cloud-ready.
Agent-agnostic AI memory MCP — never forget a conversation again
Persistent memory for Claude Code — invisible self-annotating metadata, semantic search, cross-project retrieval
Persistent memory for Claude Code & Codex CLI. Capture knowledge from sessions, compile it into a wiki, and inject the right context back into future work.
Brain-like persistent memory for AI agents. Hybrid search, knowledge graph, sleep-consolidation lifecycle, expert routing (HMoE), and a graph-Laplacian diffusion subsystem driving decay, consolidation, auto-link, and spectral retrieval. 63 MCP tools. Local, no cloud.
Local-first Personal OS + Markdown Wiki for AI agents: LLM Wiki-style knowledge graph, task claiming, and reviewable execution
Long-term memory for Claude Code and Codex that meters every token it spends
Memory module for LLMs and Agents with MCP
Persistent memory layer for Claude Code and Codex — your Obsidian vault on autopilot. Seven skills: recall, save, session notes, handoff, connections, vault health. Non-destructive, BYO vault, any taxonomy.
Cognitive memory that thinks — 7 laws, 3-tier memory, multi-agent. pip install soma-wisdom
JavaScript SDK for the Genome Evolution Protocol (GEP) -- signal extraction, gene selection, memory graph, asset store
Open-source platform for deterministic, token-aware context selection for AI agents and LLMs
The Context Operating System for AI
Your agents forget every correction. otto makes them stick.
AI research orchestration with persistent verifiable memory powered by Walrus - Sui Overflow 2026
Open source, automated memory layer that builds as you work.
AletheiaDB: The cognitive memory engine for AI agents with fact supersession.
Shared, source-backed memory for AI agents on your own Postgres. One command wires Claude Code and Codex to recall before every prompt; Cursor, Windsurf and Claude Desktop get it over MCP. Facts keep their source, supersede instead of overwrite, and clients stay apart. Keyless. Apache-2.0.
Verified experience for AI agents: independently checked, scoped, cited, and reversible.
Markdown-first persistent memory and project planning for AI agents, existing knowledgebases, and multi-agent workflows.
Give your AI coding agents memory that survives between sessions. Marrow is an MCP server providing persistent task tracking, versioned docs, and semantic code search (via a tree-sitter powered indexer) — so any agent, in any session, can pick up exactly where the last one left off.
Postgres + pgvector memory provider plugin for hermes-agent. Multi-agent storage layer with per-minion themes, async writer, no LLM in the memory hot path.
Embedded memory for TypeScript agents.
Structured conversational memory framework for LLM chatbots with semantic recall, memory grafting, and graph-based topic memory.
Project-scoped shared memory for Claude Code and Codex AI coding agents.
Salience-gated memory for Claude Code — captures what matters, forgets what doesn't, consolidates patterns while sleeping
Local-first shared memory for AI agents: persistent AI memory across Claude Code, Codex, Hermes Desktop, and any MCP client. Automatic record, recall, consolidation, forgetting, and backup. Self-hosted on Qdrant + LlamaIndex; one-command install.
Local-first checkpoints and crystals for long-running AI coding agent sessions.
Stop replaying the transcript: capsule session state + rolling memory for LLM CLIs (Claude Code, Codex). Measured: −90.3% vs no-cache / −30% vs cached replay at turn 10, reproducible bench. MIT, provider-agnostic.
Go SDK for Dakera AI agent memory — self-hosted, 88.2% LoCoMo. Vectors, hybrid search, knowledge graphs, sessions.
Long-term memory for AI agents on plain Supabase Postgres. Hybrid vector + full-text + entity recall with RRF, temporal validity, write-time dedup. You own the SQL and the rows.
Local-first nested-learning AI agent runtime with Memvid v2 memory, safe tool use, and a web workbench.
Persistent per-project memory for Claude Code, Kiro, Cursor & Codex - plain markdown committed with your code, captured prompt-free, auto-injected at session start, recalled by meaning (local semantic search, zero API). Cross-project persona, cited recall, searchable session history.
Build your agent once. Carry it everywhere. Compiles portable agent packs for Pi, Codex, Claude Code, and Cursor.
Session continuity for AI coding agents — your agent picks up where it left off. Persistent memory for Claude Code, OpenAI Codex CLI & Google Gemini CLI, sharing one local db: auto context injection, compaction handover, semantic search (KO/EN/JA), error→solution recall. 100% local, zero config, zero API cost.
Memory that abstains instead of guessing: agent memory in a local SQLite file or your own Postgres, with a verdict, confidence and provenance on every hit, and a calibrated refusal when nothing clears the threshold.
Provenance-aware memory for AI agents.
AetnaMem /ET-nuh-mem/ is agent Memory delivers fully local long-term auditable, proof-ready memory for AI Agents via a 4-tier progressive pipeline, with zero external API dependencies, it supports OpenClaw, install it via pip.
Open-source context and memory control layer for LLM applications. Inspect, manage, and selectively reuse the context your AI works with.
Governed graph-native agent memory: knowledge extraction, fusion, hybrid RAG, scoped access tokens. A push for data sovereignty.
All in one memory system for your harness
Persistent memory for Home Assistant Assist, in plain markdown. No vector DB, no embeddings — works with Anthropic Claude, OpenAI, and any LLM conversation agent.
Self-hosted long-term temporal graph memory for AI agents — MCP, REST and a dashboard in one container, with FalkorDB or Memgraph.
Place where AI Agent memories matters
An extension of your associative memory: a memo that is always with you and always yours, that brings things back the way your mind does, by association. Underneath, an index: file anything under keys, recall it by keys, plain text on your own disk.
Long-term memory engine for coding agents: two-stage retrieval + gating, hybrid BM25+vector, Ebbinghaus decay, single SQLite file, MCP-ready. Dogfooded daily: 3,300+ memories across 4.5 months.
Memory infrastructure for AI agents: store, recall by meaning or association, ingest documents, and track beliefs in one Python library.
Lightweight local memory for agents — every dedup, update, and rerank decision inspectable and tunable. 3 dedup detectors × 4 update resolvers × 3 rerank modes · 60+ parameters · 6 prompts replaceable
Stateful workflow checkpoint manager calculating state hashes and rollback points
Long-horizon conversational memory compaction & episodic anchor extractor (Mem0 / LangMem)
Semantic prompt cache deduplicator calculating Jaccard token overlap to prevent redundant LLM cost
AI_CONTEXT: a way to share durable project context between any AI agents and humans using the repository itself as the continuity layer.
Ebbinghaus forgetting curve auditor calculating retention probabilities and executing garbage collection
Agentic Soft-Token Compression (ASTC): training-free in-model context compression for long-horizon LLM agents.
Autonomous Long-Horizon Agentic Memory Pruning & Importance Attributor. Evaluates multi-turn episodic memory buffers, calculates multi-factor salience scores (recency decay, user priority pinning, emotional valence, and entity graph connectivity), and deterministically prunes low-utility noise to respect model context limits.
Biological Sleep-Cycle & Ebbinghaus Forgetting Curve Episodic-to-Semantic Memory Consolidation Engine. Decays stale transient interactions and consolidates frequent episodic events into enduring semantic user persona traits.
Autonomous Agent Tree-of-Thoughts (ToT) State Rollback Checkpointer. Allows long-running personal agents (Manus, Devin) to snapshot environment/context states, evaluate branch viability, and backtrack when an exploration path fails without restarting from scratch.
Long-Horizon Autonomous Task Checkpoint & Resumption Engine (inspired by Persistent Consumer & Enterprise Agents). Provides cryptographic event-sourcing state persistence, idempotency key deduplication, scheduled timeout heartbeats, human-in-the-loop pause-and-resume tokens, and fail-safe recovery.
Context window attention optimizer reordering retrieved documents to place critical evidence at prompt boundaries, mitigating lost-in-the-middle degradation.
Cognitive multi-tier memory kernel managing working scratchpad, sliding dialogue buffer, and episodic memories with exponential recency decay.
High-throughput semantic query cache and approximate deduplicator combining exact SHA-256 and cosine similarity threshold matching with LRU eviction.
Hybrid semantic and keyword search over Oh My Pi session transcripts
Fast semantic query cache with fuzzy term similarity and canonical intent hashing to avoid duplicate LLM invocations and cache prompt responses safely
Brain-inspired cognitive AI architecture featuring multi-layered memory (Sensory, Semantic, Episodic) and an automated "Sleep Cycle" for knowledge consolidation.