Connect your AI assistant directly to your Comp AI compliance workspace. Manage risks, tasks, vendors, policies, and more — without leaving your editor or chat interface.
Hosted server: https://mcp.trycomp.ai/mcp
No installation needed. Just point your MCP client at the hosted server.
Claude Desktop (~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"comp-ai": {
"url": "https://mcp.trycomp.ai/mcp"
}
}
}Cursor (.cursor/mcp.json in your project, or ~/.cursor/mcp.json globally):
{
"mcpServers": {
"comp-ai": {
"url": "https://mcp.trycomp.ai/mcp"
}
}
}Windsurf (~/.codeium/windsurf/mcp_config.json):
{
"mcpServers": {
"comp-ai": {
"url": "https://mcp.trycomp.ai/mcp"
}
}
}Go to Comp AI → Settings → API Keys and create a key.
Tell your AI assistant:
set-api-key <your-api-key>
That's it. You're connected.
Show me my compliance dashboard
List all open high-severity risks
Create a risk: "Third-party vendor data breach" with high likelihood and major impact
Assign task <id> to <person> and set the due date to end of month
Draft a data retention policy for our AWS setup
Show me all vendors that haven't been assessed yet
Trigger evidence collection for task <id>
Add context: our primary cloud provider is AWS with multi-region setup in us-east-1 and eu-west-1
| Tool | Description |
|---|---|
set-api-key |
Authenticate with your Comp AI API key — do this first |
compliance-dashboard |
Full overview: org info, risk/task/vendor/policy counts |
list-people |
List org members with IDs, roles, and departments |
list-risks |
List risks, filterable by severity and status |
get-risk |
Get full details of a single risk |
manage-risk |
Create or update a risk |
list-tasks |
List compliance tasks, filterable by status or assignee |
get-task |
Get full details of a single task including evidence |
update-task |
Update a task's status, assignee, or due date |
get-findings |
Get findings/gaps for a compliance task |
list-vendors |
List vendors |
manage-vendor |
Create or update a vendor |
list-policies |
List organizational policies |
manage-policy |
Create or update a policy |
list-context |
List AI context entries (Q&A pairs for policy generation) |
add-context |
Add context to help Comp AI understand your org better |
collect-evidence |
Trigger automated evidence collection for a task |
The server also exposes these resources for direct access:
comp://risks— All organizational riskscomp://vendors— All vendorscomp://policies— All policiescomp://tasks— All compliance tasks
Requires Node.js 18+.
git clone https://github.com/trycompai/comp-mcp
cd comp-mcp
npm install
npm run build
npm startThe server listens on http://localhost:3000/mcp by default.
To point your MCP client at a self-hosted instance:
{
"mcpServers": {
"comp-ai": {
"url": "http://localhost:3000/mcp"
}
}
}Environment variables:
| Variable | Default | Description |
|---|---|---|
PORT |
3000 |
Port to listen on |
TRYCOMP_API_KEY |
— | Pre-configure an API key (optional, can also use set-api-key tool) |
TRYCOMP_API_URL |
https://api.trycomp.ai/v1 |
Override the Comp AI API base URL |
Anonymous usage analytics collected via Agnost AI. No sensitive data is recorded.
MIT