Samsara MCP Is Generally Available: 40+ Read-Only Tools for Fleet Data in AI Clients

By AiCybr Editorial · Published September 25, 2026

Samsara made its hosted Model Context Protocol (MCP) server generally available on September 22, 2026, giving supported customers read-only access to live fleet and operations data from compatible AI clients. The launch includes more than 40 tools covering vehicles, drivers, safety events, hours of service and other operational records.

The service uses OAuth with each user's Samsara identity. Samsara says MCP requests inherit the same dashboard permissions, roles and tag scopes as the signed-in user, and each call remains attributable to a named user. The current tool surface is read-only: connected agents can retrieve permitted information, while account settings, messaging and assignments remain outside the GA MCP tool surface.

Samsara says the service is available to customers at no additional charge at launch. The company plans to move MCP to monthly usage-based pricing in the future; a future rate has not been published in the launch material.

Samsara MCP at a glance

What the MCP server exposes

The direct tool catalog lets an AI client query structured Samsara records. Samsara's examples include checking which drivers have hours of service remaining, finding inspection reports with unresolved defects and identifying trucks below a fuel threshold before a planned crew move.

The generally available tool surface covers fleet and asset visibility, trailer pools, driver and safety data, compliance data such as hours of service and driver vehicle inspection reports, routes, dispatch, fuel and locations.

Samsara also exposes Samsara Assistant through MCP. The launch announcement describes this as a second path for more complex questions, with the Assistant performing reasoning on Samsara's side while the external AI client provides the user-facing workflow.

This gives developers two useful integration patterns: call individual MCP tools when an agent needs structured operational records, or invoke Samsara Assistant when the task benefits from Samsara's own domain-oriented reasoning layer.

Authentication follows the Samsara user

The security model is identity-based instead of a shared API credential. The first connection opens a browser sign-in flow and authenticates the user through OAuth.

Samsara says the MCP server enforces the same permissions visible in its dashboard. An administrator can therefore access data available to that administrator, while a user restricted to a particular tag receives only data within that scope.

Existing Samsara access governance remains the authorization boundary for data returned through MCP. Routine interactive access uses each named user's identity and existing permissions instead of a shared API key or service account.

The current server is read-only. An AI client can inspect permitted operational data; write-oriented account actions remain outside the GA MCP surface. Workflows that need bulk integration or application logic continue to use Samsara's APIs and other supported automation surfaces.

Regional MCP endpoints

Samsara documents separate endpoints for its three dashboard regions:

The endpoint must match the region of the organization's Samsara dashboard. Samsara's documentation specifies that authentication is region-bound; a US-cloud organization, for example, uses the US endpoint.

Claude, Cursor, ChatGPT and Copilot integration

Samsara's launch announcement names Claude, ChatGPT, Microsoft Copilot and Cursor as examples of AI tooling that can connect to the hosted MCP service. Its developer documentation provides concrete setup instructions for Claude Desktop and Cursor and states that other clients can connect when they support remote MCP servers with OAuth sign-in.

For Cursor, the server can be registered in a project-level .cursor/mcp.json or the user's global MCP configuration. Claude Desktop organizations can add the endpoint as a custom connector, after which individual users authenticate with their Samsara identities.

Client capability still matters: an application needs compatible remote-MCP and OAuth support for this connection model. Samsara's regional endpoint and the user's existing account permissions then determine the accessible operational data.

MCP and the REST API serve different workloads

Samsara positions MCP for conversational and agent-driven access. Its REST API remains the interface for conventional integrations and moving operational data at volume.

An analyst asking an AI assistant to correlate hours-of-service records with another business system can use MCP as an interactive context layer. A production service continuously synchronizing a large operational dataset should continue to use the API and its normal integration controls.

The MCP interface can also participate in multi-system agent workflows. A compatible agent can combine permitted Samsara results with data exposed by other connected business systems, subject to the access controls of each source.

How this relates to Agent Studio

Samsara launched Agent Studio earlier in 2026 as its environment for creating and managing agents inside the Connected Operations Platform. Agent Studio includes pre-built workflows and custom agents that can automate operational tasks.

MCP exposes Samsara context to third-party AI clients and custom external agents. Samsara describes APIs, MCP, Samsara Assistant and Agent Studio as complementary surfaces over its operational data platform.

For teams already standardized on an external AI environment, MCP reduces the need to maintain a bespoke conversational connector solely to retrieve Samsara data. Teams building automations primarily inside Samsara can continue to use Agent Studio and its managed workflows.

Pricing and deployment considerations

Samsara MCP is included for customers at no additional charge at launch. Samsara says future MCP charges will be based on monthly usage, so organizations moving from evaluation to sustained agent workloads should account for a metered cost model once rates are published.

The primary deployment control is permission design. Because MCP mirrors the signed-in user's Samsara access, broad dashboard permissions also create broad read access for an AI client operating on that user's behalf. Organizations should review role and tag scope before connecting enterprise AI clients and apply their normal policies for which external AI services may receive operational data.

The read-only GA surface limits direct account-changing actions, while data returned from Samsara can still be processed by the connected AI client. Enterprise deployments therefore need to consider both Samsara authorization and the data-handling terms and controls of the selected AI platform.

Bottom line

Samsara MCP turns the company's live physical-operations data into a standardized context source for AI assistants and agents. The initial release includes more than 40 read-only tools, OAuth user identity, inherited dashboard permissions, regional hosted endpoints and no additional launch charge.

For developers, the practical value is interoperability. Existing MCP-capable environments can query Samsara without building a custom conversational integration, while the REST API remains available for high-volume application integration and Agent Studio handles Samsara-native automation. Future usage-based pricing will become an additional deployment variable when Samsara publishes rates.

Sources

• Samsara — Introducing Samsara MCP, September 22, 2026: https://www.samsara.com/blog/introducing-samsara-mcp
• Samsara Developer Documentation — MCP setup, permissions, regional endpoints and tool scope: https://developers.samsara.com/docs/mcp
• Samsara — Agent Studio launch and agentic capabilities: https://www.samsara.com/company/news/press-releases/samsara-launches-new-agentic-capabilities-to-automate-tedious-operational-tasks
• FreightWaves — independent coverage of the Samsara MCP launch: https://www.freightwaves.com/news/samsara-connects-fleet-data-with-chatgpt-claude-and-other-ai-tools