synapseai-sdk 0.1.0

Persistent memory OS for AI agents

Synapse SDK

Persistent memory OS for AI agents.

Install

pip install synapse-sdk

Quickstart

from synapse import Synapse

syn = Synapse(api_key="syn-your-key")

# Create a session for your agent
session = syn.create_session("Extract invoices from Gmail and upload to Sheets")

# Save memory after each step
syn.update_step(session, "fetch_emails", "completed", "Fetched 50 emails")
syn.save(session, "emails_fetched", "50")

# Get compressed context for your LLM
context = syn.context(session)
# → [SYNAPSE MEMORY CONTEXT]
#   Goal: Extract invoices...
#   Completed: fetch_emails
#   Next step: parse_invoices

# After any restart — resume exactly where you stopped
result = syn.resume(session)
print(result["resume_from"])  # → {"step_name": "parse_invoices", ...}

Use with any agent framework

# LangChain
from synapse import Synapse
syn = Synapse(api_key="syn-your-key")
session = syn.create_session(task)

# inject context into every LLM call
context = syn.context(session)
response = llm.invoke(context + your_prompt)

# CrewAI, AutoGen, custom — same pattern

API Reference

Method	Description
syn.create_session(task)	Create new agent session
syn.save(session, key, value)	Save memory
syn.update_step(session, step, status)	Track workflow step
syn.resume(session)	Resume after restart
syn.context(session)	Get compressed LLM context
syn.get_memory(session)	Get all session memory
syn.run_step(session, step, fn)	Auto-tracked step execution

Links

• Dashboard: https://synapse-aii.netlify.app
• API Docs: https://synapse-backend-b5k1.onrender.com/docs