Memory & context
memory layers and context management
thedotmack/claude-mem Claude‑Mem is the category leader by far and offers a token‑efficient, session‑persistent memory layer with a local worker/web UI — pick it if you want a widely adopted, general-purpose, locally runnable memory. If your team needs a specific tradeoff (reversible compression, production integration, deterministic planning, long‑term learning, or docs-grounding), choose a specialist project instead.
Which one matches your setup?
Answer any of the questions — the shortlist updates as you go. Recommendations come from the capability passports below, nothing else.
Comparison matrix
Axes are extracted from each project's docs by our review pipeline; the maturity score is computed from stars, growth and commit activity — not an opinion. Click a column to sort.
| Runs in | Models | Context | Cost to run | |||||
|---|---|---|---|---|---|---|---|---|
⭐ 101.9k +1624/7d | CLIIDECoding-agent pluginWeb app | BYOKAnthropicOpenAIGemini | Related files & memory | Configurable filters | ●●●●● | Self-hostable (local worker) | ●●●●● | Your API key for model calls; optional cloud backup to cmem.ai |
⭐ 76.5k +1318/7d | CLICoding-agent pluginWeb app | BYOKOpenAIAnthropicGeminiFixed provider | Diff + related files | Dedup & shaping | ●●●●● | Self-hostable | ●●●●● | Local software is free; LLM calls use your API key. |
⭐ 67.5k +404/7d | CLIWeb appCoding-agent plugin | BYOKOpenAI | User/session only | Multi-signal retrieval | ●●●●● | Self-hostable | ●●●●● | Your LLM/API key costs; self-host or paid cloud available |
⭐ 51.9k +2829/7d | CLIWeb appCoding-agent plugin | BYOKOpenAIAnthropicGeminiOllama / localFixed provider | Related context | Rerank & filtering | ●●●●● | Self-hostable | ●●●●● | Your LLM API key (or self-hosted models); Hindsight itself can be self-hosted |
⭐ 32.6k +436/7d | CLIWeb appCoding-agent plugin | BYOKOpenAI | Whole-graph memory | Feedback & routing | ●●●●● | Self-hostable | ●●●●● | Your LLM API key, or use Cognee Cloud (managed) |
⭐ 31.2k +3502/7d | CLIIDECoding-agent plugin | AnthropicGemini | Whole-repo analysis | BM25 & sandboxing | ●●●●● | Fully local | ●●●●● | Free, local install |
⭐ 29.7k +323/7d | CLICoding-agent pluginWeb app | BYOKOpenAIAnthropicGeminiOllama / local | Diff + related files | Confidence scoring | ●●●●● | Fully local | ●●●●● | Free to run locally; optional cloud API keys for agent LLMs. |
⭐ 28.0k +199/7d | CLIIDECoding-agent pluginCI | OpenAIAnthropicFixed provider | Project-level context | Compaction & filtering | ●●●●● | Fully local | ●●●●● | Free, open-source; external agent APIs may incur cost. |
⭐ 27.5k +119/7d | CLIIDECoding-agent plugin | BYOK | Three files only | Basic gating & checks | ●●●●● | Self-hostable | ●●●●● | Free (open-source); agent/model usage may incur costs |
⭐ 25.8k +264/7d | CLICoding-agent pluginWeb app | BYOK | Multi-file sources | Basic quality checks | ●●●●● | Self-hostable | ●●●●● | Host-dependent; may require API keys or local host resources |
⭐ 13.5k +113/7d | CLIWeb app | BYOKOllama / localOpenAIGemini | Related files | Embedding + Rerank | ●●●●● | Self-hostable | ●●●●● | No keys for demo; real usage requires your OpenRouter + DeepInfra (or other OpenAI-protocol) API keys |
⭐ 12.0k +114/7d | CLICoding-agent pluginWeb app | BYOK | Whole-memory retrieval | Smart dedup & feedback | ●●●●● | Fully local capable | ●●●●● | Cloud: API key (hosted); self-host/local available |
Capability profiles
Six axes, 0–5 each. The shape tells you the strategy: a wide hexagon is a generalist, a spike is a specialist. Showing the 8 most established — the rest are in the full catalog.
Claude-Mem pairs a Claude Code–focused persistent memory with progressive-disclosure search tools and a local worker/web UI to deliver token-efficient, session-persistent context.
Local-first, reversible context compression that both shrinks prompts and actively reduces model output tokens via a drop-in proxy and cross-agent memory.
Provides a production-ready, self-hostable memory layer with a token-efficient memory algorithm, entity linking, and agent skills for easy integration into AI assistants.
Hindsight provides a learning-focused agent memory with biomimetic data structures and a reflect operation to build mental models, going beyond simple RAG-style recall.
Runs the full memory layer (graph, vectors, sessions) on a single self-hosted Postgres-backed stack, combining vector search and evolving knowledge graphs.
Sandboxes large tool outputs and persistently indexes session events (SQLite FTS5 + BM25) so agents retrieve only relevant fragments, drastically shrinking context usage while preserving continuity across many MCP clients.
Provides a local, multi-agent persistent memory server with hybrid retrieval and confidence scoring that shares memories across any MCP/HTTP agent.
A Dolt-backed, version-controlled structured memory graph that gives coding agents dependency-aware, persistent task memory with semantic compaction for long-horizon workflows.
All repositories (41)
Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, com…
Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agen…
The Memory Layer for AI Agents - Drop-in memory infrastructure for AI agents and apps. Context that persists.…
Cognee is the open-source AI memory platform for agents. Give your AI agents persistent long-term memory with…
Context window optimization for AI coding agents. Sandboxes tool output (98% reduction), persists session memo…
#1 Persistent memory for AI coding agents based on real-world benchmarks
Persistent file-based planning for AI coding agents and long-running tasks. Crash-proof markdown plans, sessio…
Distilly — Distill how they think into reusable Skills for any Agent or Bot. Formerly Colleague Skill（原同事 Skil…
One portable memory layer for every AI agent: local-first, Markdown-native, user-owned, and self-evolving acro…
Self-evolving memory OS for LLM & AI Agents: ultra-persistent memory, hybrid-retrieval, and cross-task skill r…
Persistent memory for Claude Code, Codex and 6 other AI agents, stored as plain markdown in your Obsidian vaul…
Long-term memory runtime for AI agents — plain Markdown as the source of truth, local ranked retrieval, and an…
Knowhere extracts, parses, and outputs structured chunks ready for AI Agents and RAG.
Infrastructure for the next generation of voice agents, designed to provide universal memory. It is divided in…
Unbounded context. Memory that manages itself. One session, for life. The hippocampus for coding agents, part…
🍙 A personal AI agent & local memory hub for all AI agents, gives every AI one shared, fully controlled memo…
A memory plugin for AI coding that turns engineering experience, repository knowledge, and your way of working…
Non-destructive compression gateway for AI coding agents. Cuts token bills 25% on turn 1 to past 85% in long o…
Coding agent memory for Claude Code, Codex, Cursor and 39 more, built from the session history already on your…
ClearAI：以本体为纲的原生 DSH 插件 —— 经证据核验的研究长成可浏览的领域本体。 · ClearAI: an ontology-first DSH agent preset — evidence-checke…
Up to 71.5x fewer tokens per session on Claude Code with Obsidian + Graphify. Persistent memory, codebase know…
A self-improving skill for AI coding agents (Claude Code, Cursor, AGENTS.md): recognize a hard-won golden path…
Git-native persistent memory for AI coding agents. Implements Google OKF v0.2 with sub-300µs in-memory BM25 se…
Cut your AI coding agent's token bill on three axes: terse prose, YAGNI-first code, and tool-output compressio…
LLM-supervised persistent memory for AI agents — graph-based recall, cross-session knowledge, single binary. W…
Composable, view-based memory for DeepSeek Harness. Pluggable sources and strategies, with three-tier memory o…
Encrypted, fully offline agentic memory. One click install, GUI w/ memory map, all OS and agents. Superior mem…
A persistent, file-based memory layer for coding agents — give Claude Code, Codex & others a project brain (du…
Local long-term memory for Hermes, Codex, Claude Code, WorkBuddy and dsh agents: one SQLite store, source-back…
Persistent, provider-neutral memory for Codex, Claude Code, OpenCode, Pi, and MCP coding agents.
Turn project work into reusable knowledge — an AI-agent skill for Claude Code & Codex
CDAF (Cached Descriptive Asset Files) - open sidecar format for video so AI agents stop re-analyzing the same…
Noema long-term memory plugin for DSH: durable, inspectable agent memory with recall tools and a settings page…
dsh-sieve: context engineering & token optimization plugin for DeepSeek Harness (DSH) — tool output filtering,…
Most AI agents forget you the moment the tab closes. Constellation Engine gives them a hippocampus — a living…
Cross-agent plugin for manually excluding and editing AI conversation context, with search, filtering, hide/re…
SOUL.md governance framework for Hermes Agent — structured memory, skill management, and operational rules
Hosted alternatives
If running your own reviewer is more ops than you want, these managed services cover the same job.
The managed side of mem0: the same memory API run for you, with the vector store, extraction and retrieval hosted instead of self-operated.
Try Mem0 Platform →Hosted long-term memory for agents built on a temporal knowledge graph, so recall and fact updates come as a service rather than infrastructure you maintain.
Try Zep Cloud →The managed runtime for stateful agents from the Letta (MemGPT) team: persistent memory and agent state kept for you instead of hosting the server and its store.
Try Letta Cloud →