Connect AI Assistants to IncidentVault: MCP Server and REST API

Connect AI Assistants to IncidentVault: MCP & REST API
AI & Integrations

Connect IncidentVault to Claude, ChatGPT and Other AI Assistants

Give the AI assistant your team already uses permission to read incident and claims data, with access limited to each user’s role and controlled by your Admins.

In short

The Model Context Protocol (MCP) is an open standard that lets an AI assistant connect to other software and work with its data. IncidentVault offers an MCP connection that uses OAuth sign-in, so an assistant only gets the permissions you approve, and never more than your own role allows. Admins set up and control MCP connections inside IncidentVault. For system-to-system integrations, IncidentVault also has a REST API with scoped API keys and webhooks.

What an AI assistant can work with

IncidentVault is Copliancy’s incident reporting and claims management platform, covering incidents, claims, cases, settlements and regulatory inspections.

When you connect an assistant, you choose which kinds of IncidentVault data it can be permitted to read. Each area below has its own read permission, and the assistant only sees records your role can already see.

Incidents

Incident reports your team has logged, ready to ask about in plain language.

Claims

Claim records tied to your incidents, available to read when you grant the claims permission.

Cases

Case records your role can access, kept separate from claims by their own permission.

Locations

The locations your organization tracks, useful for questions about where activity happens.

Policies

Policy records stored in IncidentVault, readable when the policies permission is granted.

Regulatory inspections

Inspection records your team keeps in IncidentVault, under a dedicated read permission.

Reports

Report data from IncidentVault, so summaries start from the records you already have.

Dashboard and notifications

Dashboard information and notifications, each available through its own read permission.

Permissions you choose

When an assistant asks to connect, IncidentVault shows a consent screen. You decide what it can do before anything is shared.

Selected for you

Read permissions

Read permissions are preselected on the consent screen. You can review each one and clear any the assistant does not need.

Off unless you opt in

Write permissions

Write permissions stay switched off unless you choose to turn them on. Your Admins also set approval rules for actions.

Always applies

Limited to your role

Access is always limited to your own IncidentVault role. Unverified app names are flagged, and each request is single-use and expires after 10 minutes.

Admin controls

Admins manage MCP for the whole organization from inside IncidentVault.

MCP on or off

The organization’s MCP server can be disabled entirely.

Approval rules

Action tools require approval by default, and Admins set the rules.

Approved clients only

An allowlist can restrict connections to approved clients with exact callback URLs.

Documents and bulk export

Separate toggles control document access and bulk export.

Masked fields

Admins can mask fields so they are hidden from connected assistants.

Emergency shutdown

One control shuts down all MCP access and revokes every OAuth grant.

User-managed grants

Users can see the OAuth grants they have made and revoke their own.

Security activity log

A security activity log records MCP-related activity for review.

IncidentVault REST API

For integrations between systems, IncidentVault has an external REST API. Full reference material is in the IncidentVault API docs.

Access

Base URL and API keys

Requests go to https://app.incidentvault.com/api/v1/external with a Bearer API key. Keys start with iv_ and are created in Settings > API Keys.

Permissions

Scoped keys

Each key carries scoped permissions, such as read:incidents or write:claims, so an integration gets only what it needs.

Limits

Rate limits and pagination

Rate limits are set per key: 60 requests per minute by default, configurable up to 1,000, with standard rate-limit headers. List results page up to 500 records at a time.

Endpoint groups cover Incidents, Claims, Cases, Locations, Employees, Tasks, Policies, Settlements, Payments, Reserve Adjustments, Risk Checks, Regulatory Inspections, Webhooks and Reference data.

How to connect an AI assistant

Admins set up MCP connections inside IncidentVault and share the connection details with their team. The process in outline:

  1. 1

    Ask your Admin to enable MCP

    Your Admin turns on MCP for the organization, sets approval rules, and adds your assistant to the approved clients list if the allowlist is in use.

  2. 2

    Add IncidentVault in your assistant

    In an assistant that supports custom MCP connections, add IncidentVault using the connection details from your Admin. The client registers itself through OAuth.

  3. 3

    Sign in to IncidentVault

    The assistant sends you to IncidentVault to sign in. The request is single-use and expires after 10 minutes.

  4. 4

    Review the consent screen

    Check the app name, keep only the read permissions you need, and turn on write permissions only if you want them.

  5. 5

    Approve, and revoke any time

    Approve the request to finish. You can see and revoke your own grants later.

Learn more about IncidentVault

IncidentVault Platform

An overview of IncidentVault for incident reporting and claims management.

IncidentVault API Docs

Reference documentation for the IncidentVault external REST API.

Book a Demo

Arrange a walkthrough of IncidentVault with the team.

Bring Your Incident Data Into the Assistant You Already Use

Connect an AI assistant to IncidentVault with the permissions you choose, under controls your Admins set.

Frequently asked questions

MCP, the Model Context Protocol, is an open standard for connecting AI assistants to other software. IncidentVault provides an MCP connection so a compatible assistant can read the IncidentVault data you permit.
Admins set up MCP connections inside IncidentVault. Once MCP is enabled for your organization, you add IncidentVault in your assistant, sign in, and approve the permissions on the consent screen.
Only if you opt in to write permissions, which are off by default. Action tools also require approval by default under your Admin’s approval rules.
No. Access is always limited to your own IncidentVault role, and Admins can mask fields and switch off document access or bulk export.
You can see and revoke your own OAuth grants. Admins can also shut down all MCP access, which revokes every OAuth grant at once.
MCP lets a person connect an AI assistant with OAuth and the permissions they approve. The REST API is for system integrations, using iv_ API keys from Settings > API Keys, scoped permissions, rate limits and webhooks. See the API docs.
Disclaimer. The information on this page is provided for general informational purposes only and does not constitute legal or insurance advice. Incident reporting, claims handling and regulatory requirements depend on your organization’s facts and circumstances. Always consult qualified legal, insurance and risk advisers before making decisions. IncidentVault is a software platform, not a law firm or insurer.