The agentic rag, wrung out.
Multi-agent & MemorySaturday, September 26, 2026
Upcoming Goal Mode turns delegated child agents into persistent workers. A separate judge evaluates progress; the same child continues with full context until the goal is met or budget is exhausted. Moves beyond one-shot delegated prompts.
Makes multi-agent orchestration inside Hermes far more reliable for long-running or multi-step work. Reduces parent-agent babysitting.
Multi-agent & MemoryOct 27
Official plugin that lets an orchestrator route subagents to different models or Hermes profiles (fast/cheap, deep/strong, local, etc.) based on task type.
Multi-agent & MemoryNov 5
Rust-based memory system submitted to the Hermes Plugin Hackathon. Stores and searches with zero tokens, no external API, runs entirely locally. Explicitly aimed at stopping “wasting money on agent memory.”
Multi-agent & MemoryNov 5
Cleve launched on official Claude and ChatGPT marketplaces. One-click shared memory and docs across Claude, ChatGPT, and MCP-compatible agents including OpenClaw, Hermes, GrokBot, Instinct, Muse. Team workspaces save markdown docs that agents can reference. Internal use: “save to cleve” turns chat insights into durable docs.
Never miss a night
One feed for everything, or one per agent harness you run. Rebuilt every night, with full credit on every item.
