Consolidate and prune OpenClaw agent memories across sessions with a nightly dream cycle.
npx skills add https://github.com/aleph23/Natasha --skill agent-dream
Please help me install this Agent Skill. Skill: agent-dream Source: https://github.com/aleph23/Natasha/tree/main/skills/agent-dream Command: npx skills add https://github.com/aleph23/Natasha --skill agent-dream
Your OpenClaw agent often forgets context between sessions, leading to drift in memory, conflicting decisions, and misalignment with user goals. Agent Dream provides a nightly dream cycle to review, organize, prune, and reflect on memories so the agent wakes with a coherent sense of self and recent changes.
Run the setup script to auto-detect your workspace and enable the nightly dream cycle.
High-intent search queries and answers about installing and using this skill.
To preserve OpenClaw agent memory across multiple sessions, use a nightly dream cycle to consolidate, organize, and prune memories. This process maintains agent identity and learning coherence over time.
The best way to prune stale agent memory safely is using a 2-pass deletion method with automatic backups. This creates pre-dream and topic backup copies before removing stale data, ensuring full auditability and recovery.
A dream cycle works by processing agent memories through five phases: Orient, Gather, Consolidate, Prune, and Reflect. This workflow organizes recent changes, removes stale data, and enables self-reflection to maintain identity.
Yes, you can automate memory consolidation with zero configuration. The setup script auto-detects your workspace, memory structure, and sessions to quickly deploy the nightly memory management workflow across OpenClaw agents.
Yes, automated memory pruning works with any OpenClaw agent. It maintains long-term memory and identity by surfacing learnings and applying configurable scheduling to ensure safe, auditable memory management across sessions.
aaron-he-zhuManage two-layer memory with hot cache and cold storage for SEO projects.
etylsarinCapture and track agent expertise across sessions using a structured knowledge registry.
slhleosunLog experiences and generate governance-approved SOUL.md updates.
mishahaninSynthesize recent workspace learnings into structured memory files.
