Bring EQIQs into your AI chat
Hook your EQIQs account up to the AI app you already use. Then you can just ask it about the people you work with — how they think, what they need, and who they team up with best.
Quick start — paste one URL
In claude.ai and Claude Desktop, open Settings → Connectors → Add custom connector, paste the URL below and sign in once. No config file, no restart.
Your connector URL
https://adgmsnwjynqkhawhioil.supabase.co/functions/v1/mcp
Using a client without connector support? Config-file fallback
Drop one of these blocks into your client's config file and restart it.
Claude Desktop
Settings → Developer → Edit Config
claude_desktop_config.json{
"mcpServers": {
"eqiqs": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://adgmsnwjynqkhawhioil.supabase.co/functions/v1/mcp"]
}
}
}Cursor / VS Code
Settings → Tools & Integrations → MCP
~/.cursor/mcp.json · .vscode/mcp.json{
"mcpServers": {
"eqiqs": {
"url": "https://adgmsnwjynqkhawhioil.supabase.co/functions/v1/mcp",
"type": "http"
}
}
}Link your relationship intelligence to your AI
Create a free EQIQs account, then approve the one-time consent screen — that path needs no token. Running headless or in CI? Paid plans can generate a bearer access token instead.
See it working
A short walkthrough of connecting EQIQs to an AI client and asking it about real people data.
Trouble with the player? Watch on YouTube
For you
Ask your AI about people you know
For teams
See who works well together
For developers
Set it up in about two minutes
For investors
Our data plugs into any AI tool
Will this work in my app?
Check this before you start. Most setup problems are your AI app's plan rules, not the EQIQs connection.
Claude (web, desktop, mobile)
Works todayFree and Pro both work. Settings → Connectors → Add custom connector.
Cursor / VS Code / Codex CLI
Works todayPaste the config block. No plan requirement.
Terminal (any agent)
Works todaynpx -y @eqiqs/mcp-server — works with any MCP-aware runtime.
ChatGPT
Needs a paid planBusiness, Enterprise, or Edu only. Free, Plus, and Pro have no Connectors menu — that is an OpenAI plan gate, not an EQIQs problem.
Gemini / Copilot / Perplexity
Not yetNo custom MCP connector support yet. We'll list them here the day they ship it.
On ChatGPT Free, Plus, or Pro? Use Claude to see the exact same tools working — same server, same sign-in, no upgrade.
How to set it up
Three steps, about two minutes.
- 1
Copy the link below
It's one line. Copy it with the button.
- 2
Paste it in your AI app
Claude, ChatGPT, or Cursor. We show you where.
- 3
Sign in once
A sign-in box pops up. After that, just ask questions.
Your link:
https://adgmsnwjynqkhawhioil.supabase.co/functions/v1/mcpYou only sign in the first time. Nothing else to install.
Choose your AI workspace
Pick your client. Every path ends at the same server and the same one-time sign-in.
Claude
Claude on the web (easiest)
- 1In Claude, click Customize in the left sidebar, then open the Connectors tab. (There is no Connectors item under Settings — Settings only lists General, Account, Privacy, Billing, and so on.) The button below jumps straight there.
- 2Click Add custom connector.
- 3Name: type EQIQs.
- 4Remote MCP server URL: paste the link below (use the Copy button so there is no typo).
- 5Authentication: select 'Always required'. Claude usually pre-selects it and shows a 'Detected' badge — leave it. Do not pick 'None'.
- 6OAuth client: leave the recommended 'Use Anthropic's hosted client metadata' selected. If Claude shows an error after adding, come back and switch to 'No client ID — register one automatically' instead.
- 7Advanced: leave it collapsed. There is no client ID, client secret, scope, or header to enter.
- 8Click Add. On first use Claude opens an EQIQs sign-in screen — sign in and click Approve once.
On Claude Desktop the same screen is under Settings → Connectors. Custom connectors work on Claude Free, Pro, Max, Team, and Enterprise.
Claude Desktop (config file)
- 1Only needed if you want Claude Desktop to load EQIQs at startup — the Customize → Connectors path above also works inside the desktop app.
- 2Open Claude Desktop → Settings → Developer → Edit Config. That opens claude_desktop_config.json (macOS: ~/Library/Application Support/Claude/, Windows: %APPDATA%\Claude\).
- 3Paste the JSON below. If the file already has an "mcpServers" block, add only the "eqiqs" entry inside it — do not duplicate the outer braces.
- 4Save the file and fully quit and reopen Claude (closing the window is not enough).
- 5The first time a tool runs, a browser window opens — sign in to EQIQs and click Approve once.
{
"mcpServers": {
"eqiqs": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://adgmsnwjynqkhawhioil.supabase.co/functions/v1/mcp"]
}
}
}mcp-remote is the standard open-source bridge that lets a stdio-only client talk to our hosted server and handles the OAuth sign-in for you. Nothing to install first — npx fetches it on demand.
Prefer our official launcher? Swap the two args for ["-y", "@eqiqs/mcp-server"] — it's published on npm and points at this same endpoint.
Transport is streamable HTTP and authentication is OAuth 2.1 with dynamic client registration. You approve an EQIQs consent screen once and the connector acts as you, under your own row-level permissions.
Not sure what to ask it?
Twelve copy-paste prompts: interview prep, 1:1 agendas, feedback scripts, cofounder fit checks.
What-if agent simulator
How an autonomous agent works a real staffing question through the EQIQs tools. Illustrative — not live data.
list_profilesMapped 12 active employee profiles across row-level security.
get_profileExtracted 21 psychometric frameworks (MBTI, Big Five, DISC, Enneagram).
score_compatibilityCalculated interpersonal friction metrics across 4 relationship contexts.
Live tool playground
Every published tool with its request parameters and structured response shape.
request
{
"search": "jordan",
"limit": 5
}response
{
"profiles": [
{
"id": "prf_8f21",
"name": "Jordan Reyes",
"email": "jordan@example.com",
"frameworks_complete": 21
},
{
"id": "prf_4c09",
"name": "Jordana Bell",
"email": "jordana@example.com",
"frameworks_complete": 14
}
],
"count": 2
}AI archetype demos
Three ways people use the connector day to day. Scripted examples — no real profile data.
Pressure-tests founder pairings before the cap table locks.
Try a scenario
Pick a scenario above to see how the assistant answers using the underlying 21-framework profiles.
Live Response Demo
What a connected assistant returns when it calls the EQIQs tools. Illustrative example, not real profile data.
Score Jordan and Alex as a working pair and tell me how to manage them.
Across the full battery these two cover for each other's weakest side: Jordan supplies the structure Alex skips, Alex supplies the momentum Jordan defers. The friction is real but it is process friction, not values friction — which is the kind you can design around.
| Dimension | Jordan | Alex |
|---|---|---|
| Work style | Structured planner | Fast improviser |
| Decision speed | Deliberate | Decisive |
| Conflict style | Avoids friction | Debates openly |
| Feedback need | Written, specific | Verbal, immediate |
| Stress signal | Goes quiet | Adds scope |
Coach it like this
- Sequence the work before Alex starts building.
- Agree a decision deadline up front.
- Jordan needs an invitation to disagree.
- Follow every verbal note with two lines in writing.
- Silence from Jordan is the early warning, not the all-clear.
Say this in your next 1:1 with Alex
"When you move first, Jordan reads it as the decision already being made and stops pushing back. Give him the draft a day early and ask him what he'd change — you'll get the objection while it's still cheap."
Decision-support only. Never the sole basis for hiring, promotion, or termination decisions.
What it can do
Just ask in plain words. Your AI picks the right tool for you.
Read your own results
Pulls up your own 21-framework read straight from your account — no need to pick your name out of a list.
"What are my blind spots at work?"
get_my_profileSee your list of people
Shows everyone you have saved — teammates, clients, friends, or family.
"Who's on my EQIQs list?"
list_profilesRead one person
Gives the full read on one person: how they work, how they talk, what drives them, and how they act under stress.
"Tell me about Jordan."
get_profileAdd someone new
Adds a new person. You can include what you already know about them, like their personality type.
"Add Priya. She's an ENFP on the marketing team."
create_profileSave a note
Saves a note or check-in on someone, like how a 1:1 went. It shows up on their profile in EQIQs.
"Note that Priya wants more ownership of launches."
add_noteRead your notes
Reads back the notes you saved on someone, newest first.
"What have I noted about Jordan lately?"
list_notesInvite someone
Emails someone a private link to take the assessment and compare with you. Nothing about them is shared until they finish it themselves.
"Invite Jordan to compare with me."
invite_personCheck your invites
Shows who you invited and whether they have opened, started, or finished.
"Has Jordan finished yet?"
list_invitesSee saved pairs
Shows the two-person matchups you already saved, and why you saved them.
"What pairs have I saved?"
list_relationshipsScore two people
Compares two people and gives a score, what will go well, and where they may bump heads.
"How well would Jordan and Alex work together?"
score_compatibilityRead a whole team
Scores everyone in a group against everyone else in one go: how well the group works together, who holds it together, who is on the edge, which pairings need attention, and what the mix is missing. Counts as 2 calls.
"How does my product team work together?"
score_teamTake the assessment in chat
Asks the EQIQs questions one at a time right in the conversation and saves each result to your profile. Uses the same scoring as the site.
"Let's do my DISC assessment here."
start_assessmentPrep a meeting
Before a 1:1 or team meeting: how to talk with each person, and which pairings in the room need care. Counts as 2 calls.
"Prep me for tomorrow's meeting with Priya and Sam."
prep_meetingWho should lead this?
Ranks people for leading a piece of work, with the results behind each pick. A starting point for your judgment.
"Who on the product team should lead the launch?"
suggest_leadWrite the coaching read (premium)
Writes a clear, plain-language read on one person or a pair — how to work with them, what to watch for, what to do next. Counts as 5 calls because it runs a writing model.
"Write me a manager brief on Priya, focused on her first 90 days."
generate_narrative
Starter prompts
Copy one and paste it into Claude or ChatGPT once EQIQs is connected.
Your privacy, in plain words
- • Your AI only sees what you can see when you log in. It can never see other people's accounts.
- • It can read your people, add people and notes, send invites, and save your own assessment answers. It can't delete anything, and it can't touch billing or company settings.
- • You can turn it off any time from your AI app. That cuts off access right away.
- • These reports are a helping hand, not a verdict. Never use them as the only reason to hire, promote, or let someone go.
Built to cut your token bill
The naive way to give an agent psychometric context is to paste a primer on all 21 frameworks into the system prompt, then dump the whole person record on top. That is expensive on every single turn. EQIQs tools return only the fields that carry signal.
Raw profile record
~354 tokens
47 columns of JSON, per person
EQIQs tool payload
~107 tokens
signal-bearing fields only
Prompt primer avoided
~1,045 tokens
per request, every request
- • 70% smaller per profile than handing the model the raw record — measured across 200 live profiles.
- • Tool descriptions replace the framework primer, so the model learns what it can ask for without you paying for the explanation on every turn.
- • Scoring runs on our side. score_compatibility returns a number and the reasons behind it, instead of making the model derive it from two full records.
- • At 10,000 calls a month, the difference is roughly 2.5M input tokens you never send.
Narratives you can ship to customers
generate_narrative is not a raw model call with a nice prompt. Every generation is validated on our side before it reaches your agent, so what you embed in a customer-facing product stays in character.
- • Grounded, not invented. The narrative is written from that person's stored assessment data — no signal, no claim.
- • Framework names are blocked. Output that names an instrument is caught and regenerated automatically.
- • Protected characteristics, clinical language, and hire/fire verdicts are blocked the same way — the exact legal exposure that stops teams from shipping generative HR features.
- • Structured output. You get parsed sections, action items, and a
guardrailsreport on every response, not just a wall of prose. - • The decision-support disclaimer is enforced on the response, not merely requested in the prompt.
What it costs
Every tool call your AI makes counts as one API call on your Integration plan — the same allowance as the REST API, shared across both.
Sandbox
$0
500 calls / mo · 10 AI narratives
100 / hour
Growth
$99/mo
10,000 calls / mo
1,000 / hour
Scale
$299/mo
50,000 calls / mo
5,000 / hour
API Enterprise
$999/mo
100,000 calls / mo
20,000 / hour
- • Handshakes and tool discovery are free — only real tool calls are counted.
- • Every response carries live usage headers, so your agent always knows what's left.
- • generate_narrative runs a generation model, so it bills 5 calls each. Sandbox includes 10 narratives a month; paid plans draw on the full allowance.
- • Run out on Sandbox and calls pause with an upgrade message. Paid plans keep working and bill overage.
- • Track calls, per-tool breakdown, and spend in your API dashboard.
- • Embedding EQIQs in your own product? Enterprise adds committed volume, a private rate limit, and a signed data-processing agreement — email BD@EQIQs.com.
If something looks wrong
Nearly every setup hiccup is one of these eight.
Don't have an account yet?
It's free to start. Add at least one person to your account, and your AI will have something to talk about.