Build a self-updating agent memory graph with GBrain
Use GBrain to give OpenClaw or Hermes-style agents a local knowledge brain with markdown ingestion, entity links, hybrid search, scheduled consolidation, and MCP access.
npx skills add agentskillexchange/skills --skill build-a-self-updating-agent-memory-graph-with-gbrain
Use GBrain when an operator wants an agent-maintained memory layer that can ingest markdown pages, meetings, emails, voice notes, companies, people, and project facts, then wire them into a searchable knowledge graph. The workflow is to install GBrain, follow its agent installation protocol, choose the PGLite or Postgres engine, configure ingestion and scheduled jobs, then expose the brain through the CLI or MCP server for agent use.
What this skill actually does
Invoke this instead of plain vector search or scattered markdown notes when the agent needs typed relationships, timelines, recurring consolidation, identity resolution, and health checks around a persistent knowledge base. The scope boundary is operating GBrain as an agent memory and knowledge-ops layer for OpenClaw/Hermes-style agents; it is not a generic note app, hosted CRM, or broad database framework.
Inputs are a markdown knowledge base or source directories, provider keys for optional enrichment, schedule settings, engine configuration, and an agent runtime that will call the CLI or MCP operations. Outputs are indexed pages, typed links, searchable memory, scheduled consolidation jobs, health-check reports, and agent-readable operations for retrieval, enrichment, and migration.