CLS++ Integrations
Plug the CLS++ memory layer into your agent framework, editor, or workflow tool. All integrations read and write the same namespaced memory through the CLS++ brain, so memory is shared across processes, machines, and AI vendors.
Agent-framework adapters
Native memory backends shipped inside the clsplusplus package. Each is an
optional extra (pip install clsplusplus[<name>]) and is lazily imported — the
base package never requires the framework.
| Framework | Extra | Class | Plugs in as |
|---|---|---|---|
| CrewAI | clsplusplus[crewai] |
clsplusplus.integrations.crewai.CLSMemoryStorage |
ExternalMemory(storage=...) |
| LangGraph | clsplusplus[langgraph] |
clsplusplus.integrations.langgraph.CLSMemoryStore |
graph.compile(store=...) |
| LlamaIndex | clsplusplus[llamaindex] |
clsplusplus.integrations.llamaindex.CLSMemoryBlock |
Memory.from_defaults(memory_blocks=[...]) |
| AutoGen | clsplusplus[autogen] |
clsplusplus.integrations.autogen.CLSMemory |
AssistantAgent(memory=[...]) |
All four adapter constructors take the same arguments as Brain:
(user=..., api_key=None, url=None), reading CLS_API_KEY / CLS_BASE_URL
from the environment when not passed.
Editor & workflow tools (via MCP)
These connect through the CLS++ MCP server (tools recall_memories,
store_memory, who_am_i):
Claude Code and Windsurf use the same stdio MCP config shown in the
top-level README.
For a hosted, OAuth-based connector use the remote MCP endpoint
https://www.clsplusplus.com/mcp.