Quickstart
Connect any MCP-native agent to your Kiomon workspace in one command:
npx kiomon-mcp --token <your-kiomon-token> --url <your-kiomon-url> Or with environment variables:
KIOMON_TOKEN=<your-token> KIOMON_URL=<your-url> npx kiomon-mcp - Create a workspace and API token from your Kiomon dashboard.
- Set
KIOMON_URLto your workspace API base URL (shown in the dashboard). - Add the server to Claude Desktop, Cursor, or any MCP client. Check out our Claude, Cursor, and Ollama setup guides.
Memory-native tools (8)
The complete toolset built on the standard Model Context Protocol with zero vendor lock-in:
list_workspaces: Discover workspaces (and their IDs).get_workspace_context: One-shot pre-flight pack containing pinned items, Routines, Facts, and workspace stats.search_memory: Hybrid search across memories and files.get_memory: Read a memory in full, plus strength/stats and 1-hop graph relations.explore_memory_graph: Walk how memories connect: derived_from, supports, contradicts, related_to, precedes, follows.reflect_session: Batch-commit session learnings as pending drafts to the Review Inbox.draft_memory: Write a single memory as a pending draft.manage_memory: Pin, unpin, archive, restore, forget, rate, approve, reject.
Review Inbox (write guardrail)
Writes are never blindly trusted. draft_memory and
reflect_session create pending_approval drafts in
the dashboard Review Inbox; they become retrievable memories only after a
human (or an approved agent) approves them via manage_memory.
Architecture
- Capture → Clean → Connect → Remember: The automated pipeline turning raw data into a durable knowledge graph.
- Memory kinds: Episodic moments, semantic facts, procedural routines, and reference files organized visually.
- Native protocol: Standard MCP interface with direct read and graph traversal tools.
- Privacy & Security: Encrypted at rest, zero training on your data, and scoped tool access.
Support
Run the built-in diagnostic test client (npm run test:client -- --all) to verify
connectivity and tool schemas. For dedicated support, SLA inquiries, or custom deployments,
contact us.