Skip to content

skill.md

Latest commit

 

History

History
100 lines (79 loc) · 3.13 KB

skill.md

File metadata and controls

100 lines (79 loc) · 3.13 KB
Raw

Synapse Layer — Agent Skill

What Synapse Layer Does

Synapse Layer is the trust layer for the agentic internet. It gives AI agents persistent, encrypted memory with built-in governance — so agents can store decisions, recall preferences, and hand off context across sessions, models, and tools.

When to Use Synapse Layer

  • When your agent needs to remember user preferences, decisions, or constraints across sessions
  • When Agent A needs to hand off context to Agent B securely (Handover V2)
  • When you need an audit trail of every memory operation (who stored what, when)
  • When you need encrypted memory at rest (AES-256-GCM) without building your own crypto
  • When you want semantic recall — find memories by meaning, not keywords
  • When you need Trust Quotient scoring to filter reliable vs noisy memories
  • When building multi-agent systems that share context with governance

How to Install

Python SDK

pip install synapse-layer

MCP Configuration (Claude Desktop / Cursor)

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

Agent Quickstart (60 seconds)

from synapse_layer import SynapseClient

# Initialize with your API key
client = SynapseClient(api_key="sk_connect_...")

# Store a memory (encrypted server-side with AES-256-GCM)
client.store(
    content="User prefers dark mode and minimal notifications",
    agent="onboarding-v2",
    memory_type="preference"
)

# Recall by semantic similarity
memories = client.recall(
    query="user interface preferences",
    min_tq=0.7,  # Trust Quotient filter
    limit=5
)

# Delete a memory (hard-delete, LGPD-compliant)
client.delete(memory_id="mem_abc123")

Supported Frameworks

Framework Status Integration
Claude (Desktop + API) ✅ Live Native MCP
Cursor ✅ Live IDE Integration
LangChain ✅ Live Python SDK
Claude Desktop ✅ Live MCP Native

Coming Soon

Framework Integration
ChatGPT Via Synapse Proxy
CrewAI SDK Adapter
AutoGen SDK Adapter
LangGraph Orchestration
n8n Workflow
Zapier Automation

Trust & Security

  • Encryption: AES-256-GCM server-side at rest, per-operation random IV, 128-bit auth tag
  • Auth: OAuth 2.0 + PKCE S256, Bearer tokens (sk_connect_*), SHA-256 token hashing
  • Isolation: 1 user = 1 tenant = 1 private mind. Zero cross-user access.
  • Audit: Immutable three-layer lifecycle (live data, tombstone, analytics)
  • Compliance: Designed for LGPD/GDPR alignment. Hard-delete on request.
  • Trust Quotient: Per-memory confidence scoring (0–1.0)

Links