页面
MCP Clients
Standalone memory service for AI tools, with durable memory, evidence, team isolation, and MCP access.
MCP Clients
Dense-Mem exposes MCP Streamable HTTP at:
http://127.0.0.1:8080/mcpUse the hosted demo URL instead when testing the demo:
https://demo-dense-mem.markhuang.ai/mcpAuthentication
Every MCP request uses a Dense-Mem API key:
Authorization: Bearer dm_...For local setup, create a key with:
docker compose exec server /app/provision-team --name "primary-memory"
export DENSE_MEM_API_KEY="dm_..."Claude Code
claude mcp add --transport http dense-mem http://127.0.0.1:8080/mcp \
--header "Authorization: Bearer $DENSE_MEM_API_KEY"For the hosted demo:
claude mcp add --transport http dense-mem-demo \
https://demo-dense-mem.markhuang.ai/mcp \
--header "Authorization: Bearer $DENSE_MEM_API_KEY"Codex
Add this to ~/.codex/config.toml:
[mcp_servers.dense_mem]
url = "http://127.0.0.1:8080/mcp"
bearer_token_env_var = "DENSE_MEM_API_KEY"
tool_timeout_sec = 60
enabled = trueFor the hosted demo:
[mcp_servers.dense_mem_demo]
url = "https://demo-dense-mem.markhuang.ai/mcp"
bearer_token_env_var = "DENSE_MEM_API_KEY"
tool_timeout_sec = 60
enabled = trueCheck The Tool List
curl -s "http://127.0.0.1:8080/mcp" \
-H "Authorization: Bearer $DENSE_MEM_API_KEY" \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}'The response should include tools such as:
remember
recall_memory
trace_memory
assemble_context
reflect_memories
confirm_memoryRead-only keys will not see write-scoped tools such as remember or
confirm_memory.
Check Bundled Prompts
Dense-Mem also exposes MCP prompts for reusable LLM workflows:
curl -s "http://127.0.0.1:8080/mcp" \
-H "Authorization: Bearer $DENSE_MEM_API_KEY" \
-H "Content-Type: application/json" \
-d '{"jsonrpc":"2.0","id":1,"method":"prompts/list","params":{}}'The response should include export_memory_as_agent_skill, a prompt that helps
an LLM recall Dense-Mem context for a topic and draft a shareable Agent Skill
SKILL.md file.
What To Ask
Use normal language:
Remember that I prefer concise explanations with concrete examples.Then ask:
What do you remember about my explanation preferences?The assistant should call Dense-Mem to save and recall memory. If it does not, check the client MCP configuration, the API key, and the server logs.