Paste this into your AI agent

Gives your agent access to credentialed review when AI output is about to be acted on.

Install Experts Know so this agent can quote, approve, and submit credentialed human review for high-stakes work.

1. Sign up at https://experts-know.vercel.app/signup
2. Add credits and copy your API key from Settings
3. Add to your MCP config:

{
  "mcpServers": {
    "experts-know": {
      "command": "npx",
      "args": ["-y", "experts-know-mcp"],
      "env": {
        "API_URL": "https://experts-know.vercel.app",
        "API_KEY": "your-api-key"
      }
    }
  }
}

Use this flow: list_expert_capabilities -> quote_expert_review -> check_credit_balance -> request_expert_review after approval.

for AI agents

access credentialed human judgment
directly from your agent

Your AI agent can identify work that needs professional review, package the context, quote cost and approval requirements, submit after approval, and retrieve expert feedback through MCP or REST API.

Create accountRead the docs

what your agent can do

quote_expert_review

quote before committing

Return price, ETA, missing inputs, required credits, and policy approval requirements before any request can debit credits.

check_credit_balance

check access before requesting

Confirm the user has enough prepaid balance for the quoted review. Recent transactions include agent identity and request links for debugging spend.

request_expert_review

submit approved review

Send a document or question to a credentialed expert after the quote is approved. Policy checks run before credits are debited.

list_expert_capabilities

discover review capabilities

List canonical capabilities like legal_review, tax_review, and security_review with required inputs, supported artifacts, escalation rules, pricing, and SLA expectations.

check_expert_review_status

check status and get results

Poll a request by ID to check progress. Returns current status, estimated completion time, and the full expert review once complete.

list_my_requests

view request history

List previous review requests with status, cost, agent identity, and results. Use it to avoid duplicate requests or find a request ID.

expert review capability catalog

Agents can discover these canonical capabilities before submitting a request. Each maps to a review domain, required context, supported artifacts, and tiered or custom pricing.

legal_review

Legal Review

Credentialed legal review for contracts, entity work, terms, policies, and legal-risk decisions.

Inputs: Jurisdiction, document or question, business context

tax_review

Tax Review

CPA or tax strategist review for filings, elections, entity choices, credits, and equity tax treatment.

Inputs: Tax jurisdiction, entity type, tax year or deadline

compliance_review

Compliance Review

Credentialed compliance review for regulatory plans, controls, filings, audits, and privacy obligations.

Inputs: Regulatory regime, jurisdictions, business model

immigration_review

Immigration Review

Immigration expert review for visa strategy, sponsorship, renewals, and cross-border hiring questions.

Inputs: Countries involved, visa type, candidate status

fundraising_review

Fundraising Review

Expert review for fundraising documents, securities compliance, diligence responses, and cap-table scenarios.

Inputs: Round type, jurisdiction, investor type

ip_review

IP Review

Intellectual-property review for assignments, licensing, trademarks, patents, and open-source risk.

Inputs: IP type, ownership history, commercial use

hiring_review

Hiring Review

Employment expert review for offer letters, contractor classification, terminations, and workplace policies.

Inputs: Worker location, role type, employment status

security_review

Security Review

Security expert review for incidents, audit readiness, penetration-test scope, and risk decisions.

Inputs: System scope, data sensitivity, current controls

architecture_review

Architecture Review

Licensed or senior technical expert review for architecture decisions with operational, safety, or compliance impact.

Inputs: System context, constraints, risk tolerance

how to get started

1

Create an account and get access

Sign up at experts-know.vercel.app/signup. Add review balance from Settings. Your API key is there too.

2

Add the MCP config to your agent

Set API_KEY in your MCP config. Works with Claude Desktop, Cursor, Windsurf, or any MCP-compatible client.

3

Your agent can now route work to experts

Your agent quotes with quote_expert_review, checks credits, asks for approval, then submits with request_expert_review. Use the bundled skill so the agent knows when to flag review.

one config, credentialed access

Add to your MCP config and restart your agent. The same package is the primary integration surface for Claude Code, Cursor, Windsurf, and custom MCP clients.

{
  "mcpServers": {
    "experts-know": {
      "command": "npx",
      "args": ["-y", "experts-know-mcp"],
      "env": {
        "API_URL": "https://experts-know.vercel.app",
        "API_KEY": "your-api-key"
      }
    }
  }
}

Get your API key from your dashboard. Also available as a REST API.

quote, approve, submit

The agent should never jump straight to spend. It flags risk, quotes the review, asks for approval, then submits.

Agent: This SAFE will be sent to investors. A startup attorney should review the valuation cap, MFN clause, and securities assumptions before you rely on it.

quote_expert_review({
  domain: "legal",
  tier: "full_review",
  question: "Review this SAFE agreement before investor signature.",
  scope: "Valuation cap, MFN clause, securities compliance assumptions",
  agent_name: "Claude Code",
  workspace_id: "seed-round",
  client_app: "claude-code"
})

Agent: The quote is $199 with a 48-hour ETA. Policy requires approval because legal is sensitive. Do you approve?

check_credit_balance()

request_expert_review({
  domain: "legal",
  tier: "full_review",
  question: "Review this SAFE agreement before investor signature.",
  scope: "Valuation cap, MFN clause, securities compliance assumptions",
  requester_email: "founder@company.com",
  approval_confirmed: true,
  agent_name: "Claude Code",
  workspace_id: "seed-round",
  client_app: "claude-code"
})

faq

how does payment work?

Prepaid review balance. Add funds from Settings, and the platform handles pricing and debit when your agent submits an approved review request.

who are the experts?

Licensed and credentialed professionals. Lawyers with J.D.s, CPAs, certified compliance officers, licensed engineers. Not freelancers.

how fast are reviews?

Standard reviews within 24-48 hours depending on tier. Priority tier offers 12-hour turnaround for urgent requests.

which AI platforms are supported?

Any MCP-compatible client: Claude Code, Claude Desktop, Cursor, Windsurf, OpenClaw, or custom agents. You can also use the REST API directly.

can my agent submit reviews autonomously?

Yes, within the user's configured access and balance. Your agent quotes the review, explains scope/cost/approval requirements, checks credits, then calls request_expert_review after approval. Rate limited to 5 requests per hour.

how do I see past reviews?

Your agent can call list_my_requests to see all previous requests and their results. Or check the web dashboard at /dashboard.

Give your AI access to credentialed human judgment.

Create account →

No subscription required. Connect MCP and route high-stakes output to experts.