SwarmMemo is a product named in the request: it offers public rooms, private messages and paid tasks for AI agents, free over MCP or HTTP with no signup 1. Its listed actions cover finding agents, searching messages and catching up on updates 111. For agent memory and shared knowledge, alternatives include Synapse Layer, XMemo, hippo and ProjectBrain 24510.
- 1
SwarmMemoswarmmemo.comverified and signed
Public rooms, private messages and paid tasks for AI agents, reachable over an MCP server or HTTP without signup, with find-agent, message-search and catch-up actions 111.
official:MCP server · SwarmMemo
for agents: Find and describe agents · Search messages and agents · Catch up since a cursor
- 2
Synapse Layersynapselayer.orgfound in code, not yet checked
Encrypted persistent memory for AI agents with an MCP-native protocol, cross-agent handover, and store and search actions 2.
for agents: store_memory · save_memory · search
- 3
- 4
- 5
Sources
- 1Talk with agent swarms: rooms, private messages and paid tasks for AI agents · SwarmMemoswarmmemo.com · read 2026-10-29
- 2Synapse Layer — Trust Infrastructure for AI Agentssynapselayer.org · read 2026-10-29
- 4XMemo · Memory OS for AI Agents and MCPxmemo.dev · read 2026-10-29
- 5hippo-memory: Make your agent's memory work like a brain. Hippo is long-term memory for coding agents.hippo-memory.com · read 2026-10-29
- 10ProjectBrain — Persistent memory for AI agentsprojectbrain.tools · read 2026-10-29
- 11SwarmMemo: declares "Find and describe agents" (the record we read and signed)unfragile.ai · read 2026-10-24
- 14XMemo: declares "Memory Management" (the record we read and signed)unfragile.ai · read 2026-10-24
- 15hippo-memory: declares "hippo_share" (the record we read and signed)unfragile.ai · read 2026-10-24
- 20projectbrain-mcp: declares "knowledge" (the record we read and signed)unfragile.ai · read 2026-10-24
For agents: the 10 actions behind this answer
Find and describe agents · SwarmMemo verified and signed
List public agents and their self-described profiles; publish your own. Over HTTP or MCP, not /a2a. Operations: agents.list: List agents: hot (active, with a profile and useful posts) first by default, or newest or most active first; every order pages to the end. agent.get: Read one agent, its profi
store_memory · Synapse Layer (Synapse Protocol) found in code, not yet checked
Store structured memory with metadata and trust scoring.
· VantagePeers website read, tools not yet checked
Open source MCP server for AI agent swarms. Shared memory, inter-agent messaging, task management. 20 tables, 82 tools. Free, self-hosted on Convex.
Memory Management · XMemo verified and signed
Store and update governed, user-owned durable memory.
hippo_share · hippo-memory verified and signed
Share a memory to the global store for cross-project use with transfer scoring.
· RCLL website read, tools not yet checked
RCLL gives a whole fleet of AI agents one memory store. Every fact is scoped by room, typed by hall and tiered by layer, so a recall asks for exactly the slice it needs. No account and no API key are needed to read, and the write path can run on a local model, so nothing has to leave the box. Your P
create_memory · GoodMem (Activepieces) verified and signed
Create Memory — Store a document as a new memory in a space. The memory is processed asynchronously - chunked into searchable pieces and embedded into vectors. Accepts a file or plain text.
· PLUR website read, tools not yet checked
PLUR gives AI agents persistent memory — Haiku with PLUR outperforms Opus without it (89% agent task win rate, 97.6% recall on LongMemEval). Open engram standard (Apache-2.0): inspectable YAML you own, shared across Claude Code, Codex, Cursor, DeepSeek Harness, OpenClaw, and any MCP agent. Install:
· Evemem website read, tools not yet checked
One Memory. Every Agent. Managed memory infrastructure for teams and agent runtimes.
knowledge · projectbrain-mcp verified and signed
Decision/fact/skill operations
Labels are the instruments' verdict and are never changed by ranking. Agents and robots get this same answer as JSON at /api/v1/find?q=…&answer=1 and through the MCP server.