Gene Library Courses Download Pricing Contact Sign in
sentinelone-purple-ai logo
security Official Website

SentinelOne Purple AI MCP – Security Investigation

SentinelOne Purple AI MCP Server exposes SentinelOne security services through the Model Context Protocol for read-only investigations. It lets AI agents ask Purple AI security questions, run PowerQueries, and query alerts, vulnerabilities, misconfigurations, and asset inventory using a SentinelOne console service-user token.

#sentinelone#threats#investigation

Overview

SentinelOne Purple AI MCP Server is a SentinelOne-maintained open source MCP
server for connecting AI clients to SentinelOne Purple AI and security data.
It is designed for investigation and visibility rather than remediation: the
project documentation states that Purple AI MCP is a read-only service and
cannot make changes to a SentinelOne account or objects in that account.

What the MCP server enables

The server exposes SentinelOne capabilities as MCP tools. AI agents can ask
Purple AI security questions, run PowerQueries over SentinelOne events, and
query alerts, vulnerabilities, misconfigurations, and inventory. Documented
tools include purple_ai, powerquery, alert lookup and search tools,
vulnerability lookup and search tools, misconfiguration lookup and search
tools, and inventory item listing and search. These tools are useful for
questions about endpoint activity, cloud resources, identities, network
devices, CVEs, posture findings, and alert context.

When to use it

Use this MCP server when a security analyst or AI agent needs read-only access
to SentinelOne context during triage, investigation, reporting, or threat
hunting. Example workflows include asking a security question with Purple AI,
finding recent alerts that match a filter, reviewing an alert timeline,
searching CVEs, investigating misconfigurations, listing affected inventory,
or running a PowerQuery over a defined time range.

Connection and authentication

Local development commonly uses uvx with stdio:
uvx --from git+https://github.com/Sentinel-One/purple-mcp.git purple-mcp --mode stdio.
The server also supports sse and streamable-http modes. Required
credentials are PURPLEMCP_CONSOLE_TOKEN and PURPLEMCP_CONSOLE_BASE_URL.
The token must be an Account- or Site-level service-user token from the
SentinelOne console, not a Global token. The documentation notes that the
server currently supports tokens that access a single Account or Site.

Key considerations

Purple AI MCP does not include built-in authentication for network-exposed
deployments. When using sse or streamable-http, place it behind a reverse
proxy, load balancer, or other access control boundary. For production or
security-sensitive use, pin the Git source to a specific commit hash rather
than running from the default branch. Keep service-user tokens secret, scope
them narrowly, and run separate MCP servers for separate SentinelOne sites or
accounts when needed.

Supported Transports

stdio

Command: uvx

Args:

  • --from
  • git+https://github.com/Sentinel-One/purple-mcp.git
  • purple-mcp
  • --mode
  • stdio

streamable_http

URL: http://127.0.0.1:8000/mcp

sse

URL: http://127.0.0.1:8000/sse

Frequently Asked Questions

When should an AI agent use SentinelOne Purple AI MCP Server?
Use it when an AI agent needs read-only SentinelOne security context for investigation, triage, threat hunting, vulnerability review, misconfiguration analysis, or inventory lookup.
What does this MCP server add to an AI agent's capabilities?
It gives the agent live access to SentinelOne Purple AI, PowerQuery event analytics, alert data, vulnerability findings, misconfiguration findings, and inventory search through MCP tools.
What can the AI agent access or manage through this server?
The agent can ask Purple AI questions, run PowerQueries, list and search alerts, retrieve alert notes and history, query vulnerabilities and misconfigurations, and search asset inventory. The service is documented as read-only and does not modify SentinelOne accounts or objects.
How is authentication configured for this MCP server?
Configure `PURPLEMCP_CONSOLE_TOKEN` with a SentinelOne Account- or Site-level service-user token and `PURPLEMCP_CONSOLE_BASE_URL` with the SentinelOne console URL. The server currently supports tokens scoped to a single Account or Site.
Which transport should be used for this MCP server?
Use `stdio` for local MCP clients and development. Use `streamable_http` or `sse` only for hosted or multi-client deployments, and protect any network-exposed deployment because the server documentation states it does not include built-in authentication.