









































Works out-of-box with your favorite AI agents

HSI Donesafe EHS is a no-code EHS platform, covering incidents, audits, training, and compliance. Makini's MCP server brings HSI Donesafe EHS data and workflows to AI agents directly, with no custom integration to build.
Have any questions? We’re here to help You
Yes, HSI Donesafe is the Donesafe platform under its current owner: HSI, the US safety training and EHS software company Health & Safety Institute, acquired Donesafe on February 4, 2020. Donesafe was founded in Sydney, Australia in 2013 as a cloud safety and compliance platform, and the product kept its name and no code approach after the deal. HSI now markets it as HSI Donesafe, with 60 plus configurable modules and about 3 million users.
No code in HSI Donesafe means administrators build and change modules without developers: forms are assembled from 30 plus field types, workflows are drawn with drag and drop and conditional logic, and dashboards are configured per role. A company can rename statuses, add fields or build a new module in hours. For an AI agent this cuts both ways: the data model is yours, so field names and statuses differ between Donesafe customers.
Yes, HSI Donesafe has a secure RESTful API, plus SFTP staging hosts for bulk data exchange, and the vendor lists integrations with HR, payroll and ERP systems and with IoT devices. Single sign on is supported through SAML 2.0, Okta, ADFS 3.0 and OAuth. The API is a customer tool; an MCP agent does not call it directly, since Makini handles the Donesafe side once you have authorized a connection through Makini Link.
Yes, Makini runs an MCP server for HSI Donesafe EHS. Sign in to Donesafe through Makini Link, paste the MCP endpoint Makini issues into Claude, ChatGPT, Cursor or another MCP client, and call get_connection_details to see which Donesafe objects and fields the connection exposes. Because every Donesafe instance is configured differently, that call matters more here than for most systems; EHS records appear only where the connection supports them.
HSI Donesafe groups its modules into Health and Safety, Risk and Compliance, Quality and Supplier, and Environmental and Sustainability. Incident Management and Hazard Management sit in the first group with Safety Meetings and PPE; Contractor Management and Vendor and Supplier Management sit in the third, and Permit to Work, Visitor Management and Learning Management are available as well. Chemical and SDS Management covers hazardous substances, and customers can add modules they build themselves.
An AI agent can summarize HSI Donesafe incidents and hazards by site when the connection exposes those records: ask for incidents in the last 90 days grouped by site and type, hazards still open past their review date, or sites with rising near miss counts. Because Donesafe modules are customer configured, check the field names with get_connection_details first and tell the agent which status means closed in your instance before trusting the totals.
Treat writes to HSI Donesafe as an exception: start the agent on reading incidents, hazards and actions, and let people log and close records themselves, since incident reports and action closures are compliance evidence. Where the connection and the connecting user's permissions allow writes, limit them to drafting a record for a supervisor to review. The agent should never attest to a training completion, sign a permit or mark a corrective action verified.
HSI Donesafe is a cloud service hosted on AWS, with data sovereignty options so an organization's data can stay in its region, so no firewall change is needed for Makini to reach it. IT should still decide which Donesafe user authorizes the connection in Makini Link, because the MCP endpoint is scoped to that login and logged. Allowlisting Makini's fixed IP addresses or a VPN applies only to on premises systems, not to Donesafe.