Synapse Layer — Zero Knowledge Memory
Install the Synapse secure-memory MCP package when you want Zero-Knowledge Context and Neural Handover as the trust layer for agent recall.
Overview
Synapse secure-memory is an MCP server for the Build phase that implements Zero-Knowledge Context and Neural Handover for encrypted agent memory.
What is this MCP server?
- Zero-Knowledge Context and Neural Handover branding for cross-session agent continuity
- AES-256-GCM encryption with optional SYNAPSE_ENCRYPTION_KEY (auto-generated if omitted)
- SYNAPSE_PRIVACY_EPSILON differential privacy default 0.5
- stdio MCP via uvx on PyPI package synapse-layer v1.0.7
- SYNAPSE_AGENT_ID distinguishes claude-desktop, custom agents, and multi-agent setups
- Manifest version 1.0.7
- Package identifier synapse-layer on PyPI with stdio transport
- Four documented environment variables including secret encryption key
Community signal: 9 GitHub stars.
What problem does it solve?
Handing context between agents or sessions usually means copying chats or unsafe local files that expose secrets.
Who is it for?
Builders standardizing on Synapse’s trust-layer narrative for Claude Desktop or custom MCP agents.
Skip if: Quick prototypes that do not need encryption keys or differential privacy tuning.
What do I get? / Deliverables
Registering synapse-secure-memory lets agents read and write a encrypted memory layer with privacy epsilon and per-agent IDs through standard MCP stdio.
- MCP-accessible secure memory with encryption key management
- Privacy-tuned recall via SYNAPSE_PRIVACY_EPSILON
- Isolated memory namespaces per SYNAPSE_AGENT_ID
Recommended MCP Servers
Journey fit
Secure memory is configured during build when you choose how agents persist state alongside your repo and deployment stack. Listed under agent-tooling because it is an MCP trust layer for memory, not a ship-time test or launch SEO tool.
How it compares
Sibling MCP package to synapse-layer with the same PyPI runtime—memory trust layer, not a payment or browser QA server.
Common Questions / FAQ
Who is synapse-secure-memory for?
Developers and solo founders who run MCP agents and want branded zero-knowledge memory instead of rolling their own encrypted store.
When should I use synapse-secure-memory?
During agent stack design when you need durable context handover between tools, sessions, or agent personas.
How do I add synapse-secure-memory to my agent?
Configure MCP stdio with uvx for synapse-layer from PyPI, set SYNAPSE_AGENT_ID, and provide SYNAPSE_ENCRYPTION_KEY if you want a fixed key.