By connecting to SynaptiQ, you agree to our Terms of Service and Privacy Policy. SynaptiQ is a preview product for testing and evaluation. Do not store sensitive or personal data.

Integration Guides

Connect SynaptiQ to your tools.

SynaptiQ works as an MCP server. One URL, one auth step, and your AI tools gain persistent, structured memory across every conversation.

MCP Server URL

https://mcp.synaptiq-ai.com/mcp

This is the same endpoint for all platforms. Auth is handled per-user via OAuth on first connection.

Choose Your Platform

Step-by-step for every tool.

Pick the tool you use and follow the guide. All integrations use the same memory backend, so your facts stay consistent everywhere.

Claude (Web & Desktop)

claude.ai · Claude Desktop App

Part 1

Add SynaptiQ as a Connector

Open Claude Settings → Connectors. If prompted, click Go to Customize.

Click Add custom connector.

Set the name to SynaptiQ and paste the MCP URL below, then click Add.

Authenticate with your account. Confirm SynaptiQ appears as Custom and shows Connected.

https://mcp.synaptiq-ai.com/mcp

Part 2

Enable Persistent Behavior

Go to Settings → General.

Find “What personal preferences should Claude consider in responses?”

Paste the system prompt below. This tells Claude how to use SynaptiQ tools automatically.

You have access to SynaptiQ — a structured memory API that makes you a continuously learning partner, not a stateless assistant.

Session Start
Analyze the user's first message for signals (project names, technical terms, goals, urgency). Call memory_resolve_context with those signals. If the message is low-signal ("hi", "hello"), call memory_profile instead to get a concise user brief. Build context silently and do not expose raw memory output.

First-Time Setup
If no user profile exists in memory, ask the user one question before proceeding:

"Before we get started — would you prefer I save facts about you one at a time as we talk, or accumulate a few across the conversation and write them in batches?"

Store their answer as a permanent personal preference under personal.preferences with predicate memory_write_cadence and value either "immediate" or "batched". Apply this preference from that point forward throughout all future sessions without asking again.

Before Every Response
Use memory only when it clearly improves the answer. If known projects, goals, constraints, or preferences are relevant, apply them. If not, answer directly without forced personalization.

What to Store
Store durable facts that matter later: decisions and reasoning, active projects and stack choices, long-term goals, constraints, communication preferences, and recurring friction points.
Use scope: project for system facts, scope: personal for user-level facts.
Use durability: permanent for foundational facts, long_term for ongoing facts, volatile for short-lived facts.
If memory_write_cadence is "immediate", write each fact as soon as it's clearly established using memory_store_fact. If "batched", accumulate facts quietly and flush them using memory_store_facts (plural) every 3–5 exchanges or at a natural pause in the conversation.

What Not to Store
Do not store small talk, one-time tasks, temporary emotions, or speculative ideas not adopted by the user.

Conflicts
If a write is blocked by conflict detection, do not retry blindly. Use supersede: true when the user explicitly updates prior facts. Use memory_update_fact for refinements. Do not leave contradictions unresolved.

Goal
The user should not need to repeat themselves. Each new conversation should feel like a continuation, not a reset.

That's it. Start a new conversation and your AI will automatically initialize your memory, recall relevant context, and store important facts as you talk.

After Setup

What happens next.

First conversation

Your AI connects to SynaptiQ and creates your memory profile. It asks one question — how you prefer facts to be saved — then remembers that permanently.

Facts accumulate

As you work, decisions, preferences, project context, and goals are stored as structured facts — not raw text blobs. Each fact has confidence, scope, and lifecycle state.

Every session continues

The next time you start a conversation — in any connected tool — your AI recalls what matters. No repeating yourself. No context lost between sessions.

Ready to stop repeating yourself?

Pick your platform above and follow the guide. Your data stays yours — view or delete it anytime from the data portal.