[FEATURE] New project: Mnemoverse memory agent example (memory_agents)

Feature Description

Add a runnable memory agent example that uses Mnemoverse, a hosted memory service for AI agents, through its MIT-licensed Python SDK (pip install mnemoverse). The agent writes facts from a conversation, reads the relevant ones back in a later session, and reports whether a recalled memory helped or misled, so the service re-ranks what comes back next.

License split: the Python SDK and the MCP server (@mnemoverse/mcp-memory-server) are MIT. The memory engine is a hosted service with a free tier, so running the example needs a free Mnemoverse API key. SDK docs: https://mnemoverse.com/docs/api/python-sdk

Disclosure: I work on Mnemoverse.

Target Project

mnemoverse_memory_agent

Project Directory

memory_agents

Motivation

The Memory Agents section shows capture and recall with Memori, Agno, and Weaviate Engram. This example adds the step after recall: the agent tells the memory layer how a recalled memory worked out, and that report changes what the next recall returns. It gives readers a small, complete write, read, and feedback loop to study.

Proposed Solution

• Location: memory_agents/mnemoverse_memory_agent/, one self-contained project in one PR, linked with Closes #<this issue>.
• Model calls through Nebius Token Factory, following the existing repository pattern.
• Memory through the mnemoverse Python SDK: write after a turn, read before answering, and send feedback when the user confirms or corrects an answer.
• Use case: a project assistant that keeps decisions and preferences across sessions.
• README.md based on .github/README_TEMPLATE.md, a pyproject.toml, and a .env.example with placeholder keys only.
• Clear handling when the Mnemoverse key is missing or the service cannot be reached.
• One catalog line in the root README.md under Memory Agents.

User Impact

Developers get a runnable example of closing the memory loop with outcome feedback, placed next to the other memory examples so the approaches can be compared side by side.

Alternatives Considered

An MCP version using @mnemoverse/mcp-memory-server under mcp_ai_agents/. The SDK version fits memory_agents/ better, because the example is about the memory loop rather than tool discovery.

Screenshots/Mockups

A screenshot or GIF will be added to the project's assets folder with the PR, as the PR template asks.

Implementation Checklist

• I have searched for similar feature requests
• I have provided a detailed description of the feature
• I have explained the motivation and user impact
• I have considered alternative solutions