LeandroPG19/cuba-memorys — Persistent memory MCP server for AI agents — Rust, 19 tools, knowledge graph, Hebbian learning, epis…

Description and use

**Brief summary**
LeandroPG19 / cuba‑memorys is a Rust‑based persistent‑memory MCP (Model Context Protocol) server that equips AI agents with a rich episodic knowledge graph, Hebbian learning, contradiction detection, prospective triggers, and Bayesian calibration. It ships as a zero‑config Docker image and bundles 19 ready‑to‑use tools for connecting agents to external data and services.

**Value proposition**
The project gives developers a drop‑in back‑end that turns raw tool outputs into a structured, self‑updating memory store, enabling agents to reason over past interactions, detect inconsistencies, and generate context‑aware prompts. By exposing a standard MCP API (plus an SDK and CLI), it removes the need to hand‑craft bespoke memory layers, accelerating the creation of “tool‑aware” assistants and simplifying the rollout of Model Context Protocol servers.

**Practical adoption path**

| Step | Action | Reason |
|------|--------|--------|
| 1️⃣ Clone / pull the Docker image | docker run leandropg19/cuba-memorys | Immediate, zero‑config sandbox for evaluation. |
| 2️⃣ Connect a prototype agent | Point the agent’s MCP client to http://localhost:… and use the provided SDK (Rust/Python bindings) to store/retrieve memories. | Validates API compatibility and data flow. |
| 3️⃣ Extend with custom tools | Register additional tool‑handlers via the MCP register_tool endpoint or the CLI. | Demonstrates the “19 tools” extensibility and integrates domain‑specific services. |
| 4️⃣ Tune learning & calibration | Adjust Hebbian learning rates, Bayesian priors, and trigger thresholds via the REST config endpoint. | Aligns the memory behavior with the target use case (e.g., customer support vs. research). |
| 5️⃣ Harden for production | Add TLS termination, persistent volume mounts, and monitoring (Prometheus exporter is built‑in). | Meets security, observability, and durability requirements. |

**Production readiness**

• **Maturity:** Medium – the codebase is functional and actively updated (last commit 2026‑07‑13) with a clean Docker deployment, making it suitable for prototypes and internal tools.
• **Stability signals:** 26 stars, 3 forks, and a modest but focused Rust community; the API surface is stable, but the project lacks a formal CI/CD pipeline and extensive integration tests.
• **Operational considerations:** Verify the open‑source license, perform a security audit of the Docker image, and pin dependency versions (Rust crates) before using it in a production environment. With those checks, the server can be promoted to production for workloads where occasional memory drift is acceptable (e.g., research assistants, internal chatbots).

In short, cuba‑memorys offers a ready‑made, extensible memory layer for AI agents, with a straightforward Docker‑first onboarding path; after a brief security and dependency review, it can move from prototype to production for many internal‑facing AI services.

Problem

LeandroPG19/cuba-memorys helps connect AI assistants to real tools and data through a standard protocol.

Outcome

Medium: useful for prototypes or internal workflows, with dependency and maintenance checks before production.

Workflow

• connect AI agents to tools
• ship Model Context Protocol servers
• standardize integrations

Evidence

Persistent memory MCP server for AI agents — Rust, 19 tools, knowledge graph, Hebbian learning, episodic memory, contradiction detection, prospective triggers, Bayesian calibration, zero-config Docker setup.

Source