Synapse Layer Docs

First Memory in 3 Minutes

pip install synapse-layer

from synapse_layer import SynapseClient

client = SynapseClient(api_key="sk_connect_...")
client.store(content="User prefers dark mode", agent_id="onboarding-v2", type="MANUAL")
memories = client.recall(query="user preferences", reason="Loading user context for session.")

curl -X POST https://forge.synapselayer.org/api/v1/memory/commit \
  -H "x-connect-token: $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"content": "User prefers dark mode", "agent_id": "my-agent"}'

How It Works

Synapse Layer provides an AES-256-GCM encrypted memory layer accessible via MCP protocol and REST API. Memories are encrypted server-side before storage in PostgreSQL.

Search and recall operate over indexed metadata and derived fields. Memory content remains encrypted at rest (AES-256-GCM) and is decrypted only for authorized responses. Conflict resolution runs via Trust Quotient™ scoring on metadata.

Note: Trust Quotient™ is a per-memory confidence signal (0.0–1.0) combining recency, consistency, confidence, source, and integrity. It is not a security certification or external trust score.

Security — Tested Controls

Every security control below is implemented and production-tested. Implemented controls are validated through internal evidence-based engineering. External audit is not claimed unless explicitly stated. Public components are available on GitHub. Features marked 🔜 are planned.

Core API

Authentication

REST API (Forge): x-connect-token: $TOKEN — REST API uses Connect Tokens via x-connect-token header. Tokens are generated through Synapse Connect™ (OAuth 2.0 + PKCE S256).

MCP Server (/api/mcp): Authorization: Bearer sk_connect_... — MCP Server uses API keys (sk_connect_*) via Authorization Bearer header. Keys are generated from the Forge Dashboard.

Site Capture Gateway (/api/v1/capture): Authorization: Bearer sk_connect_... — Site capture gateway uses Connect Tokens (sk_connect_*) via Authorization Bearer header.

MCP Server Configuration

Synapse Layer is MCP-native. Connect any MCP-compatible client (Claude Desktop, Cursor, custom agents) to the memory server. Tools exposed: recall, save_to_synapse, process_text, search, health_check, initialize_context, save_memory, store_memory, recall_memory, list_memories, memory_feedback, neural_handover, slo_report (13 tools via tools/list).

Framework agnostic — memories stored via one MCP-compatible client can be recalled from any other using the same API key.

{
  "mcpServers": {
    "synapse-layer": {
      "url": "https://forge.synapselayer.org/api/mcp",
      "headers": {
        "Authorization": "Bearer sk_connect_..."
      }
    }
  }
}

🚀 Roadmap Synapse Layer

Synapse Layer builds the infrastructure for persistent memory in AI.

While traditional systems merely retrieve data (RAG), Synapse Layer lets agents genuinely remember, evolve, and maintain context over time.

RAG retrieves. Synapse remembers.

✅ Delivered

• AES-256-GCM encryption at rest (per-operation random IV)
• MCP JSON-RPC native protocol
• REST API + official Python SDK + TypeScript SDK
• Framework agnostic (Claude, Cursor, LangChain via MCP)
• OAuth 2.0 + PKCE S256 (Synapse Connect™)
• SHA-256 content deduplication
• Cross-agent memory tagging via agent_id
• PII detection (11 regex patterns, auto-redaction)
• Differential privacy noise injection

🔄 In Progress

• Secure Agent Handover (V2) — SDK-local JWT packaging works; server-side persistence in progress
• Consent screen UI (Synapse Connect™ UX refinement)
• Semantic recall via pgvector HNSW indexing
• Memory ranking optimization (semantic + recency + scope)
• Multi-agent memory orchestration
• Observability layer (latency, recall precision, traceability)
• Automated security testing pipeline
• Audit logs and traceable memory access
• Token lifecycle hardening

🔜 Planned

• Multi-tenant memory isolation
• Advanced PII redaction layer
• Regional data residency
• Enterprise SLAs and dedicated monitoring
• Long-term context compression
• Autonomous memory prioritization
• SOC 2 Type I