All insights extracted from relevant context engineering and memory papers
See the codeREADME
momo-research
Welcome to momo-research. This is a public research log for Context Engineering and Persistent Memory for AI agents.
This repository serves as a archive of insights from papers focused on:
- Context engineering
- Long-term and persistent memory for agents
- Memory architectures
- Unified organizational memory
- Agent workflows and memory-aware system design
Our goal is to translate complex technical research into actionable insights that help developers, builders, and researchers understand how AI agents can leverage memory more effectively.
What You’ll Find Here
We will continuously update this repository with:
Paper Summaries
Concise breakdowns of key papers, including:
- Google’s whitepaper on Context Engineering
- Manus’s blog on Context Engineering for AI Agents
- Chroma's blog on Context Rot
- The Complexity Trap: Simple Observation Masking Is as Efficient as LLM Summarization for Agent Context Management
- Google's Evo-Memory: Benchmarking LLM Agent Test-time Learning with Self-Evolving Memory
- Multi-Agent Collaboration via Evolving Orchestration
- CodeAct: Executable Code Actions Elicit Better LLM Agents
- Recursive Language Models
- DARWIN GÖDEL MACHINE: OPEN-ENDED EVOLUTION OF SELF-IMPROVING AGENTS
- AUTOMATED DESIGN OF AGENTIC SYSTEMS
- OMNI-EPIC: OPEN-ENDEDNESS VIA MODELS OF HUMAN NOTIONS OF INTERESTINGNESS WITH ENVIRONMENTS PROGRAMMED IN CODE
- Ontology Embedding: A Survey of Methods, Applications and Resources
Let me know if there are other memory related papers.
Why This Repo Exists
We believe the next generation of AI agents won’t be defined by model size,
but by their ability to remember, integrate, and act on information across many apps over time.
This repository documents our process as we:
- Study foundational research
- Extract the core principles of context engineering
- Apply these insights to build real memory-as-a-tool modules for modern applications
Our goal is to push forward the understanding and implementation of persistent memory systems and openly share what we learn along the way.
Implementation Notes
As we develop memory-as-a-tool modules for apps such as Slack, Gmail, and Linear, we will share:
- Architecture sketches
- Design rationales
- Lessons learned
We are also building a lightweight Memory Playground where developers can test these modules, inspect how memory is stored and retrieved, and experiment with different context-engineering strategies in a real environment.
Stay Updated
Follow progress here:
- Product website: https://usemomo.com
- X: https://x.com/cailynyongyong
Contributing
If you're researching similar topics or want to collaborate, feel free to:
- Open an issue
- Submit a pull request
- Reach out directly
Contributors
All insights extracted from relevant context engineering and memory papers
See the codeREADME
momo-research
Welcome to momo-research. This is a public research log for Context Engineering and Persistent Memory for AI agents.
This repository serves as a archive of insights from papers focused on:
- Context engineering
- Long-term and persistent memory for agents
- Memory architectures
- Unified organizational memory
- Agent workflows and memory-aware system design
Our goal is to translate complex technical research into actionable insights that help developers, builders, and researchers understand how AI agents can leverage memory more effectively.
What You’ll Find Here
We will continuously update this repository with:
Paper Summaries
Concise breakdowns of key papers, including:
- Google’s whitepaper on Context Engineering
- Manus’s blog on Context Engineering for AI Agents
- Chroma's blog on Context Rot
- The Complexity Trap: Simple Observation Masking Is as Efficient as LLM Summarization for Agent Context Management
- Google's Evo-Memory: Benchmarking LLM Agent Test-time Learning with Self-Evolving Memory
- Multi-Agent Collaboration via Evolving Orchestration
- CodeAct: Executable Code Actions Elicit Better LLM Agents
- Recursive Language Models
- DARWIN GÖDEL MACHINE: OPEN-ENDED EVOLUTION OF SELF-IMPROVING AGENTS
- AUTOMATED DESIGN OF AGENTIC SYSTEMS
- OMNI-EPIC: OPEN-ENDEDNESS VIA MODELS OF HUMAN NOTIONS OF INTERESTINGNESS WITH ENVIRONMENTS PROGRAMMED IN CODE
- Ontology Embedding: A Survey of Methods, Applications and Resources
Let me know if there are other memory related papers.
Why This Repo Exists
We believe the next generation of AI agents won’t be defined by model size,
but by their ability to remember, integrate, and act on information across many apps over time.
This repository documents our process as we:
- Study foundational research
- Extract the core principles of context engineering
- Apply these insights to build real memory-as-a-tool modules for modern applications
Our goal is to push forward the understanding and implementation of persistent memory systems and openly share what we learn along the way.
Implementation Notes
As we develop memory-as-a-tool modules for apps such as Slack, Gmail, and Linear, we will share:
- Architecture sketches
- Design rationales
- Lessons learned
We are also building a lightweight Memory Playground where developers can test these modules, inspect how memory is stored and retrieved, and experiment with different context-engineering strategies in a real environment.
Stay Updated
Follow progress here:
- Product website: https://usemomo.com
- X: https://x.com/cailynyongyong
Contributing
If you're researching similar topics or want to collaborate, feel free to:
- Open an issue
- Submit a pull request
- Reach out directly