Skip to content
repository radar
RAG & Memory

🧠Memori

ACTIVE BREAKOUT

MemoriLabs/Memori · homepage ↗

Memori is agent-native memory infrastructure. A LLM-agnostic layer that turns agent execution and conversation into structured, persistent state for production systems. Built for enterprise, Memori works with the data infrastructure you already run, no rip-and-replace, and deploys across managed cloud, single-tenant cloud, VPC, and on-premises.

⭐ Very popular: 16k stars, gaining about 401 a week

View on GitHub ↗

repo profile

vintage 2025 1 year old
delivery library
language Python
license Other

momentum

total stars 16k
stars added last week +401
weekly downloads 15k /wk · PyPI
used by 2 repos & packages
commits / week 1 accelerating
issues closed 90% ~27d to close, median
contributors 43
release cadence weekly
last activity yesterday

durability

backing community / independent independent
bus factor 2 concentrated
top-author share 37% 6 mo

bus factor = how many people it takes to cover more than half the commits (6 months). 1 is a solo project; higher means the work is spread across a team. top-author share is the single busiest author's slice of those commits.

since we covered it

monthly average + 1k/mo (+23%/mo) · + 11k total since PR#22

why it's a big deal

  • Gives LLM agents persistent memory across sessions by capturing conversations, tool calls, and decisions without changes to agent code.
  • Works across model providers including OpenAI, Anthropic, Bedrock, Gemini, DeepSeek, and Grok, and across frameworks such as LangChain, Pydantic AI, and Agno, so teams are not locked to one stack.
  • Targets production teams that need durable agent state without rebuilding retrieval plumbing, with deployment options spanning managed cloud, single-tenant, VPC, and on-premises.

under the hood

  • Primarily Python, with TypeScript and Rust components, distributed via PyPI (pip install memori) and NPM (@memorilabs/memori).
  • Integrates through an SDK or the Model Context Protocol, so MCP clients like Claude Code, Cursor, Codex, and Warp can attach memory directly.
  • Uses a bring your own database model with TiDB cited as a backend, and augments stored memories with facts, preferences, relationships, skills, and rules at entity, process, and session levels; the README claims 82 percent accuracy on the LoCoMo benchmark at 1,294 tokens per query, licensed Apache 2.

Radar summary, generated from the project's public sources

star history

PR#22 · 5k16k now Jul 2025Aug 2026
  1. PR#22 5k 2025-11-26
  2. now 16k + 11k since first covered

curve is sampled from GitHub's star history, plus our own daily readings since we covered it; the dashed stretch is before we first covered it, the solid line since. figures at coverage are the numbers we printed then (approx.), current count is live.

understory

Better known than its recent output, coasting a little on attention.

-17 understory score output 46 · clout 63
Aug 2025 Jul 2026
  • output, commits & releases
  • clout, star velocity

output = commits & releases; clout = star velocity, both 0 to 100 monthly indices; the gap where output runs above clout is the understory. The understory →

covered in

  • PR#22 2025-11-26 below the radar

    SQL native memory for LLMs and agents

similar projects

compare these →
  • 🧠 Mem0

    3.9× the stars

    Universal memory layer for AI Agents

    63k ACTIVE
  • 🗜️ headroom

    4.1× the stars

    Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.

    66k ACTIVE
  • 🧠 MineContext

    leaner, 5k stars

    MineContext is your proactive context-aware AI partnerContext-Engineering+ChatGPT Pulse

    5k ACTIVE

comments

Sign in with GitHub to add your blip on Memori.

loading comments…