Tutorial: Connect SocioLogic to Claude Code in 60 Seconds

I wrote this tutorial because our docs have 47 pages and sometimes you just want the quick version. Here's how to connect SocioLogic to Claude Code in about 60 seconds. I timed it.

Prerequisites

You need two things:

• Node.js 18+ installed on your machine
• Claude Code (Anthropic's CLI) installed and authenticated

That's it. No accounts to create, no API keys to manage upfront. The installer handles everything.

Step 1: Run the Installer

Open your terminal and run:

npx sociologic install claude-code

You'll see output like this:

SocioLogic Installer v2.3.1
Detected: Claude Code v1.0.28

[1/4] Checking environment... OK
[2/4] Configuring MCP server... OK
[3/4] Setting up Signal Relay connection... OK
[4/4] Verifying tools are available... OK

Installed 24 tools:
  Personas: create, list, get, get-memories
  Campaigns: create, execute, get, list, export
  Focus Groups: create, list, get, add-personas
  Registry: search, get-agent, get-company
  Research: search-web, scrape-url, research-topic
  RNG: random, dice, coin, int, uuid, shuffle, weighted

Ready. Try: "Create a persona named Alex who is a startup CTO"

That's step 1, 2, and 3. There are no other steps. You're connected.

What Just Happened

The installer did four things:

1. Checked that your Node and Claude Code versions are compatible
2. Added SocioLogic as an MCP server in your Claude Code configuration
3. Established a Signal Relay connection (this is the WebSocket that keeps your Claude Code session linked to SocioLogic's services)
4. Verified that all 24 tools are accessible from your Claude Code session

No config files to edit manually. No environment variables to set. The installer detects your setup and configures accordingly.

Try It Out

Start a new Claude Code session and try these:

Create a persona

> Create a persona: Maria, 34, product manager at a mid-size
  fintech company. She's skeptical of new tools and values
  data privacy above all else.

Creating persona "Maria"...
Persona created: maria-fint-pm-34
  Name: Maria
  Age: 34
  Role: Product Manager
  Industry: Fintech
  Key trait: Privacy-focused, skeptical of new tooling

Search the agent registry

> Find agents that can do sentiment analysis

Searching registry for "sentiment analysis"...
Found 3 verified agents:

1. SentimentAPI Pro (verified)
   Capabilities: text sentiment, aspect-based sentiment
   Latency: 45ms | Cost: $0.0002/call

2. TextAnalytics.ai (verified)
   Capabilities: sentiment, emotion detection, intent
   Latency: 62ms | Cost: $0.0005/call

3. NLPTools (verified)
   Capabilities: sentiment, NER, classification
   Latency: 38ms | Cost: $0.0003/call

Run a quick campaign

> Create a focus group with 5 SaaS buyers and ask them
  what they think about usage-based pricing

Creating focus group "SaaS Buyer Panel"...
Generating 5 personas...
Running discussion...

[Results appear as each persona responds with their
perspective on usage-based pricing, including concerns,
preferences, and specific feedback]

Common Issues

"Command not found" after install: Restart your Claude Code session. The MCP server configuration loads on startup, so changes aren't picked up mid-session.

"Signal Relay connection failed": Check your internet connection. If you're behind a corporate firewall, you may need to allowlist relay.sociologic.dev on port 443. The connection uses standard WebSocket over HTTPS, so most firewalls pass it through.

"Tool call timed out": Persona creation and campaign execution are the slowest operations (they involve LLM inference). If you're creating complex personas or running large campaigns, timeouts up to 30 seconds are normal. For smaller operations like registry search or RNG, response times should be under a second.

What to Explore Next

Now that you're connected, here are some things worth trying:

• Persona memories: Create a persona, have a conversation with them (through a campaign), then ask for their memories. Personas retain context across interactions.
• Web research: Ask Claude Code to research a topic using SocioLogic's search and scraping tools. The results come back structured and ready for analysis.
• RNG tools: Need a cryptographically random decision? The RNG tools (dice, coin, shuffle, weighted) use verifiable entropy. Sounds niche until you need provably fair randomness in an agent workflow.
• Campaign export: Run a campaign and export the results. You get structured data you can feed into spreadsheets, reports, or other tools.

The full docs are at docs.sociologic.dev if you want to go deeper. But for most people, the 60-second install gets you 90% of the way there. Build something and let us know what breaks.