CLOUDM0N
Repository intelligence

moorcheh-ai/memanto

GitHub

An open-source, MIT-licensed Memory Agent that runs alongside AI agent fleets to curate, consolidate, reconcile, and manage long-term agent memory across frameworks and platforms.

CLOUDM0N decision
REVIEW BEFORE ADOPTION
Trust REVIEW · 80/100
Good fit if

Developers and teams running fleets of AI coding agents across multiple platforms (like Claude Code, Cursor, Windsurf, Cline, etc.) who need a unified, shared long-term memory layer.

Watch out for

Requires writing and maintaining custom YAML policy files under ~/.memanto/policies/ to configure retention and forgetting rules.

Practical intelligence

What matters before you adopt it

Problem it solves

The fragmented, unmanaged state of AI agent memory across different developer silos, where agents independently write to proprietary vector databases but cannot reconcile contradictions, expire old preferences, or share knowledge across vendors.

Best for
Developers and teams running fleets of AI coding agents across multiple platforms (like Claude Code, Cursor, Windsurf, Cline, etc.) who need a unified, shared long-term memory layer.
Organizations requiring strict data privacy who want to run long-term memory management entirely on-premises using local LLMs.
Main trade-offs
Requires writing and maintaining custom YAML policy files under ~/.memanto/policies/ to configure retention and forgetting rules.
Cross-project benchmark scores (like LongMemEval and LoCoMo) are highly sensitive to reader/judge model choices, meaning they are directional rather than absolute metrics.
Why it stands out
Zero-dependency developer setup with a single pip install containing a built-in retrieval and semantic engine.
Highly portable and open architecture using the Open Knowledge Format (plain Markdown files) to prevent vendor lock-in.
Supports 100% offline, on-premise operation using local Docker and Ollama with no external API calls.
Trust & CVEs

Security evidence without the noise

Trust remains a decision signal; CVEs and scanner evidence explain what is driving the risk.

Security findings
2
Trivy + CLOUDM0N
Critical
0
High
0
Medium
2
Low
0
View trust evidence & security findings▼
Why this score
CLOUDM0NCLOUDM0N scanner found behavior that requires manual review.
TRIVYTrivy found no critical/high vulnerability, secret, or high-risk misconfiguration signal.
OPENSSFOpenSSF Scorecard evidence is unavailable for this scan.
CLOUDM0N findings
MEDIUM
Dynamic code execution pattern detected. 5 sample match(es) found.
MEDIUM
The project can spawn operating-system processes; review command construction and input handling. 5 sample match(es) found.
Architecture from code10 modules · 4 edges▼
Structural evidence

Modules and dependency edges extracted from repository code. This is code evidence, not README inference.

Code files
310
Modules
10
Dependency edges
4
Core modules
examples
104 files
integrations
57 files
memanto
84 files
scripts
1 files
sdks
19 files
Dependency flow
examples→memanto
integrations→memanto
scripts→memanto
Detected languages
Python · TypeScript · JavaScript
Detected frameworks
Anthropic SDK · FastAPI · LangGraph · NumPy · OpenAI SDK · Pydantic · Requests · Streamlit · pandas
Architecture evidence details
flowchart TD
    %% memanto — high-level architecture (DRAFT, refine me)
    n0["examples · examples · 104 files"]
    n1["integrations · 57 files"]
    n2["memanto · 84 files"]
    n3["scripts · scripts · 1 file"]
    n4["sdks · 19 files"]
    n5["tests · tests · 45 files"]
    n0 --> n2
    n1 --> n2
    n3 --> n2
    n5 --> n2
    class n0 docs
    class n3 infra
    class n5 test
    classDef docs fill:#9d7660,color:#ffffff,stroke:#7c5d4c
    classDef infra fill:#b35c00,color:#ffffff,stroke:#8f4a00
    classDef test fill:#499894,color:#ffffff,stroke:#397975
Evidence, security & integrations▼
Integrations
Claude CodeCursorCodexWindsurfAntigravityGemini CLIClineContinue
Security notes
• On-premises mode guarantees that zero data, code snippets, or search contexts leave the host infrastructure.
• Memory visibility is restricted by default using isolated, agent-specific namespaces so production agents do not access experimental contexts.
• Forgetting is designed as an auditable state transition where expired memories remain readable in history as [EXPIRED] until permanently purged via a delete command.
Still unknown
• The README does not provide specific hardware sizing recommendations, baseline memory requirements, or CPU allocations needed to host the Docker semantic engine.
• Does not state the exact database technology used inside the local Docker containers to store and index raw records.
• No step-by-step documentation on how to configure and run the REST API behind a production reverse proxy is provided in the text.
Adoption guidance▼
Adopt if
+ You run a fleet of diverse AI agents and want them to share a consistent, managed knowledge base that updates and cleans itself autonomously.
+ You require strict intellectual property and data sovereignty guarantees where agent memory remains 100% on-premises.
Avoid if
– You are looking for a simple, passive vector database SDK to persist logs rather than an active, policy-driven curation agent.
How it works & getting started▼
How it works
1.The user installs Memanto locally and connects it to their active AI coding agents.
2.Memanto extracts durable facts, decisions, and preferences from ephemeral chat transcripts.
3.On a daily schedule, Memanto runs background loops to consolidate duplicates and flag conflicting beliefs for human review.
4.When an agent is triggered, Memanto intercepts and briefs it with only the minimal, relevant slice of memory.
5.The memory state is continuously maintained as portable Markdown files in the Open Knowledge Format.
Getting started
→ Install the tool using: pip install memanto
→ To run the local API server and web UI, run: memanto serve
→ Open the interactive browser dashboard with: memanto ui
→ Connect to your preferred IDE tool using a command like: memanto connect claude-code
Agent handoff
Use with any agent
JSON API
Alternatives

Nearby repositories worth comparing before adoption.

Compare top options →
TencentDB-Agent-Memory
TencentCloud/TencentDB-Agent-Memory
70
Fit

TencentDB Agent Memory is a team-level memory hub for AI Agents — turning conversations, docs, and code into four reusable memory assets (Chat Memory, Skill, LLM-Wiki, Code-Graph) that are governed, shared, and equipped across agents and frameworks.

Trust REVIEW · 0
Compare →
Waggle-mcp
Abhigyan-Shekhar/Waggle-mcp
41
Fit

MCP server for external memory layer for AI agents + more . Download from pypi , and get started

Trust REVIEW · 80
Compare →
agentops
AgentOps-AI/agentops
61
Fit

Python SDK for AI agent monitoring, LLM cost tracking, benchmarking, and more. Integrates with most LLMs and agent frameworks including CrewAI, Agno, OpenAI Agents SDK, Langchain, Autogen, AG2, and CamelAI

Trust REVIEW · 80
Compare →