Product: 1099Policy MCP

Your AI agent runs risk checks. Verifies coverage. Audits COIs.

1099Policy as a Model Context Protocol server. Claude, Cursor, ChatGPT, and any MCP-compatible client can call our tools through natural language. Connect once. Your agent stays in your stack.

Connect your agentRead the spec
What it does

Production-grade tools. In your agent's hands.

The same contractors, jobs, quotes, assignments, policies, certificates, and invoices your developers reach through the REST API — exposed through the MCP spec. Hosted, typed, audited.

Standardized tool surface

checkCoverage, getQuote, bindAssignment, and getCertificate — plus the rest of the catalog. Typed inputs, structured outputs, MCP-compliant.

Any MCP client

Claude Desktop, Claude Code, Cursor, VS Code, ChatGPT, Continue, Cline, Zed, and any other MCP-compatible host. One URL, every agent.

Read-only by default + audit log

OAuth sign-in grants read-only access; write tools require a write-scoped API key. Every tool call lands in the account audit trail.

Setup

Add 1099Policy to your agent in 30 seconds.

One URL. The server is hosted by 1099Policy: point your client at https://mcp.1099policy.com/mcp, approve the browser sign-in, and the tools appear in your agent. Nothing to install, no daemon to maintain.

Configuration is per-client. Add the same URL to Claude Desktop, Claude Code, Cursor, VS Code, ChatGPT, and any other MCP-compatible host.

View installation guide →
Tool catalog

Every product capability, typed and discoverable.

The MCP server exposes the same surface as the REST API: verify coverage, quote and bind assignments, pull certificates, review contractor-provided COIs, list active policies, and pull audit trails.

Each tool ships with typed schemas and structured outputs your agent can reason over.

Browse all tools →
Conversation flow

Natural language → coverage verified in seconds.

Does our HVAC sub in PA have active WC and GL for Saturday's install? becomes a typed tool call with the right contractor, work_state, and effective_date.

If the request is ambiguous, the tool returns a structured error your agent can reason over.

Read prompt patterns →
Multi-tenant safety

Read-only by default. Every call audited.

Signing in through OAuth grants read-only access: an agent you authorize in the browser can look things up but can't change anything. Tools that bind or cancel coverage require a write-scoped API key.

Every tool call lands in the account audit trail, retrievable any time with getAuditTrail.

See token management →
Trusted by

The teams that staff, dispatch, and pay contractors.

Wieden+KennedyBBDODDBGolin KetchumOctagonCreatorIQ
FAQ

Common questions before you connect.

Any MCP-compatible client, including Claude Desktop, Claude Code, Cursor, VS Code, ChatGPT, Continue, Cline, Zed, and local agent hosts. The same URL works everywhere.

Coverage checks, contractor and job management, quotes, assignment bind / extend / cancel, certificate retrieval and COI review, enrollment links, invoices, and the account audit trail — every tool with typed schemas.

Signing in through OAuth grants read-only access, and your client registers itself automatically — no key to create or paste. For headless agents, present a 1099Policy API key as a bearer token, created in the dashboard under Developers → API keys.

Only with a write-scoped API key. OAuth sign-ins and read-only keys can verify and inspect coverage without changing anything.

The setup guide and tool reference live in the 1099Policy docs.

Add 1099Policy to your agent in one command.

Point your client at the hosted server, sign in from your browser, and start asking your agent about coverage.

$claude mcp add --transport http 1099policy https://mcp.1099policy.com/mcp⌘C to copy