openclaw - 💡(How to fix) Fix [Feature]: Request for attribution: Dreaming memory system design derived from Nuwa (shierduan/Nuwa) [1 participants]

Summary

I request adding a clear attribution in OpenClaw’s documentation to credit Nuwa (shierduan/Nuwa) as the original source of the core concept and design behind the Dreaming memory system.

Problem to solve

OpenClaw’s Dreaming memory system shares the exact same core definition and architecture as Nuwa’s memory-dreamer module, which I open-sourced on December 6, 2025 — months before the feature was released in OpenClaw. There is currently no attribution to the original source. Original commit proving the timeline:https://github.com/shierduan/Nuwa/commit/6db55de28397834f5761c68a6c02969f2d5bf7e6

Proposed solution

Add a simple line in the README, acknowledgments, or feature documentation crediting Nuwa for inspiring the Dreaming memory system’s core design. Example: The Dreaming memory system concept was inspired by Nuwa (https://github.com/shierduan/Nuwa).

Alternatives considered

No response

Impact

Affected users/systems/channels: OpenClaw users, contributors, and the broader open-source AI memory community. Severity: Low (non-functional issue, but impacts open-source attribution norms and recognition of the original work). Frequency: Always (ongoing, permanent without attribution). Consequence: Without attribution, the community cannot trace the origin of the design idea, and the original work remains uncredited. Adding attribution aligns with standard open-source practices and respects community contributions.

Evidence/examples

The architectural and conceptual overlap goes far beyond a general "memory" feature, mirroring the exact pipeline implemented in Nuwa months earlier:

1. Identical Three-Stage Lifecycle:

Nuwa (Dec 2025): Implemented a 3-step loop: 1. 遗忘低价值记忆 (Forgetting low-value memory) -> 2. 压缩相似记忆为摘要 (Compressing similar memory clusters) -> 3. 铭刻高价值记忆 (Engraving high-value memory to long-term storage).

OpenClaw Dreaming: Uses the exact same logical pipeline mapped to sleep stages: Light (filtering/cleaning) -> REM (pattern extraction/clustering) -> Deep (engraving to MEMORY.md).

No response