Feature request: discuss optional ClawMem-backed persistent memory integration

[EricZequan]

Summary

I'd like to discuss an optional integration path between DeepPavlov and ClawMem for dialog systems or chatbot-style applications that need durable memory across sessions.

This is not a request to add persistence-from-scratch to every DeepPavlov model.
The request is specifically about whether DeepPavlov should support or document a ClawMem-backed optional integration for users who want:

• durable memory across sessions
• auditable memory/workspace boundaries
• GitHub-compatible issue/comment-backed memory spaces
• optional shared memory across agents or operators

Why this seems feasible

From the current repository/docs:

• DeepPavlov describes itself as a modular and configuration-driven framework.
• The README explicitly says users can build a model from components.
• The CLI includes riseapi, which runs a REST API server.
• The project explicitly covers dialog systems and chatbots.

That suggests a reasonable integration path could be an optional memory-aware component, connector, or docs example rather than a mandatory core redesign.

Important caveat

ClawMem is oriented around durable memory artifacts and GitHub-compatible repository / issue / comment primitives.
DeepPavlov is oriented around model pipelines, config-driven assembly, and task-specific components.

Because of that, the likely MVP is not "replace DeepPavlov's internal architecture with ClawMem semantics".
Instead, a narrower path seems more realistic, such as:

• an optional component / service adapter for chatbot or dialog pipelines
• an external package with a documented DeepPavlov integration example
• a docs-level reference architecture for DeepPavlov + ClawMem deployment

What I want to clarify

Would maintainers be open to one of these directions?

1. External package + docs example

   • integration lives outside core
   • DeepPavlov documents the recommended wiring pattern

2. First-party optional integration surface

   • a bounded optional component / connector in DeepPavlov

3. Not a good fit for this repo

   • if ClawMem belongs in a separate integration repo instead of DeepPavlov itself, that would be useful to know early

Minimum viable integration scope

If this is worth pursuing, the smallest useful scope seems to be:

• focus on chatbot/dialog-style workloads rather than all model types
• keep the integration optional
• preserve existing default behavior for users who do not need persistent memory
• start with docs/example or an external package unless maintainers prefer first-party support

If this direction sounds reasonable, I can turn it into a concrete implementation brief.