Compress conversations into durable memory observations across sessions.
npx skills add https://github.com/monasprox/openclaw-memory --skill total-recall-monasprox
Please help me install this Agent Skill. Skill: total-recall Source: https://github.com/monasprox/openclaw-memory/tree/main Command: npx skills add https://github.com/monasprox/openclaw-memory --skill total-recall-monasprox
The Total Recall skill provides autonomous memory augmentation for agents by compressing conversations into prioritised notes, preserving cross-session context without a database, and surfacing important items for recall.
Install the skill and run the observer to begin compressing conversations into memory.
High-intent search queries and answers about installing and using this skill.
Autonomous memory for agents is achieved by compressing conversations into prioritized notes written directly to a markdown file. This pipeline operates across sessions without requiring a database to preserve context.
The Dream Cycle is a nightly consolidation layer that processes and organizes accumulated session observations. It maintains searchable memory by running semantic hooks to ensure consolidated notes remain accessible for future agent recall.
Yes, jq and curl are required dependencies to run memory compression. The skill relies on these tools to invoke LLMs via OpenRouter or OpenAI-compatible endpoints and process conversation transcripts into durable observations.
OpenRouter endpoints are fully supported for autonomous AI memory compression. The system uses curl to invoke any OpenAI-compatible endpoint, allowing pluggable Ambient Intelligence Engine connectors to process and compress transcripts.
Auditing cross-session memory observations is handled by reviewing the flat markdown file where compressed notes are stored. This transparency allows direct debugging of the multi-layer pipeline including the Observer and Reflector outputs.
File-based memory observations should be avoided when your agents require real-time concurrent write access or complex relational querying. This flat markdown approach is designed for durable, sequential cross-session context rather than high-concurrency environments.
Mesh-LLMConnect agent tools and OpenAI clients to a mesh-llm node.
unclehowellStore durable facts and synchronize memory across agent sessions.
arisngStores issues and dependencies in a git-backed store for persistent cross-session memory.
eyadsibaiImplement persistent memory systems for AI agents using knowledge graphs.
