Every agent you run. One memory.
Remnic is one memory store behind every tool. State a preference in one agent and every other agent reads it, because they all read and write the same plain markdown files on your disk. Native plugins for the tools that support them, generic MCP and HTTP for everything else.
Host integrations
Each integration reads and writes the same memory on your machine. The recall and observe cadence depends on how deeply the host plugs in.
Claude Code
Native hooks plus MCP. Recall on every prompt, observe on every tool use.
Codex CLI
Native hooks, MCP, and a memory extension. Recall on every prompt, observe on every tool use.
ChatGPT (developer mode)
MCP plus OAuth 2.1 on your own server. On-demand recall inside the composer.
Pi Coding Agent
Native extension, MCP, and compaction coordination. Recall and observe every turn.
Hermes
Python MemoryProvider. Recall on every LLM call, observe every turn.
OpenClaw
Native memory-slot plugin. Recall every session, observe every response.
Cursor and MCP clients
Generic MCP config for Cursor, Cline, Windsurf, Amp, and Replit. On-demand recall.
WeClone
Bring your WeClone clone data in as memory with the WeClone importer.
lossless-claw
Import lossless-claw session archives into your memory store.
Ingest sources
Beyond agent hosts, Remnic can pull in the things you record and the documents you already keep.
Generic MCP and HTTP API
No native plugin? Remnic speaks MCP over both stdio and HTTP, so any MCP-compatible client can recall, store, and search through the same set of tools. Point the client at the running server:
export REMNIC_AUTH_TOKEN="$(openssl rand -base64 32)"
npx remnic-server --host 127.0.0.1 --port 4318 \
--auth-token "$REMNIC_AUTH_TOKEN"
MCP clients connect at http://localhost:4318/mcp with a
bearer token. Custom agent frameworks can skip MCP and hit the REST API
directly:
curl -X POST http://localhost:4318/engram/v1/recall \
-H "Authorization: Bearer $REMNIC_AUTH_TOKEN" \
-H "Content-Type: application/json" \
-d '{"query":"authentication patterns","topK":8,"mode":"minimal"}'Per-client setup snippets live in the connector setup guide, and the full tool surface is in the HTTP and MCP API reference.