Skip to content

Can our privacy team use OneTrust in Claude?

OneTrust in Claude lets a privacy team check DSAR stages, consent records and vendor assessments. Allow the reads and restrict the bulk deletes.

Uday Gajavalli 6 min read
A privacy lead asking Claude which data subject requests are still in progress, with a list of OneTrust requests and their workflow stages beside the chat

Why a privacy team wants OneTrust in Claude

Emily Carter runs privacy at a software company with customers in the US and Europe. Her week starts with the same questions. Which data subject requests are still in progress? Which subtasks are due this week? Did this customer withdraw marketing consent, and when? Which vendor assessments are still open, with risks unresolved? Each answer sits in OneTrust, in a different module. OneTrust in Claude puts those lookups in the chat where she drafts the reply.

MCP (Model Context Protocol) is the standard way an AI assistant calls tools in other apps. Elaichi's onetrust connector is the path from Claude to a live OneTrust account. It signs in with a OneTrust OAuth client credential (a client ID and secret that an app uses instead of a person's login). It carries more than 600 tools, from DSAR requests and consent records to assessments and the data inventory.

Most of those tools are not for a privacy desk. Some delete DSAR records in bulk, withdraw consent or purge consent data subjects. The work is choosing the handful of reads Emily's team needs and keeping the rest out of reach.

Which OneTrust scopes should the client credential get?

Give the credential read scopes only, one per module the team reads. OneTrust's quick start says client credentials are created under Global Settings, Access Management, Client Credentials, on the Client Credentials tab. You select the scopes when you create it (developer.onetrust.com, checked November 2026).

OneTrust's scopes page says tokens for a credential are authorized only for its assigned scopes (developer.onetrust.com, checked November 2026). For this playbook, pick DSAR_READ, CONSENT_READ, ASSESSMENT_READ and INVENTORY_READ. OneTrust's scope tables put data subject details under CONSENT_READ, assessment lists under ASSESSMENT_READ and inventory lists under INVENTORY_READ. The same page advises a _READ scope for any integration that only retrieves data.

The scope split matters most for requests. OneTrust lists its Delete Requests endpoint under DSAR_WRITE, not under DSAR_READ. DSAR_READ is not purely a read, though. Its table includes Reprocess System Subtask, a PUT. CONSENT_READ lists Delete Purposes from Data Subject, a DELETE. So treat the scope as the first wall and the Elaichi rules as the second. If the team later needs to move a request's stage, give that job its own credential and role rather than widening this one. OneTrust's token reference also caps an account at 500 client credentials (developer.onetrust.com, checked November 2026).

Seven read tools cover Emily's questions, and each one narrows to a stage, a person or a module. list_all_onetrust_datasubject_requestqueues lists DSAR requests and filters by stage, created date and modified date. onetrust_requestqueues_searches_search finds a request by the data subject's name, email address or request ID. It is a POST, but Elaichi tiers it as a read. Asking for one stage at a time keeps the answer short on a busy queue.

Deadlines come from the subtasks. list_all_onetrust_requestqueue_subtasks lists the subtasks of one request, with each subtask's approver and deadline. So "what is due this week" is two steps: filter requests by stage, then read their subtasks.

For consent, list_all_onetrust_datasubjects_basic_details returns a data subject's created date, last transaction date and data elements. list_all_onetrust_datasubjects_preferences returns their purposes and status in one Preference Center. That answers "did this customer opt out, and of what".

For vendors and the data map, list_all_onetrust_assessments returns each assessment's status, deadline and open risk count. It filters by template type, and VENDOR is Third-Party Risk Management. list_all_onetrust_inventory_inventories lists vendors, assets, processing activities or entities.

Which OneTrust tools delete records or change a request?

The riskiest OneTrust tool for a privacy team is onetrust_requestqueues_bulkdeletes_bulk_update. Its name says update, yet its description says it permanently deletes DSAR requests in bulk. The subtasks, attachments and Results Summaries go with them, and the description calls this irreversible. It can select by date range, stage, country or request type. It deletes up to 50 per call and reports the rest as requestsPending, so a model could simply call it again.

Consent has its own deletes. onetrust_consent_datasubjects_bulk_delete permanently deletes up to 999 consent data subjects per call. onetrust_consent_datasubject_purposes_bulk_delete removes one purpose from all data subjects. onetrust_datasubjects_preferences_bulk_delete withdraws a person's consent for every purpose in a Preference Center. onetrust_datasubjects_merges_bulk_update merges duplicates and deletes the extra records afterward. onetrust_consentmanager_v_2_datasubject_profiles_bulk_delete removes up to 100 purposes from one data subject, or one purpose from up to 100.

Kind Tools For the privacy role
Request reads list_all_onetrust_datasubject_requestqueues, onetrust_requestqueues_searches_search, list_all_onetrust_requestqueue_subtasks Allow
Consent reads list_all_onetrust_datasubjects_basic_details, list_all_onetrust_datasubjects_preferences Allow
Assessment and inventory reads list_all_onetrust_assessments, list_all_onetrust_inventory_inventories Allow
Request status onetrust_requestqueue_movestages_bulk_update, onetrust_requestqueue_pausedeadlines_bulk_update, onetrust_datasubject_requestqueues_bulk_update Restrict
Consent changes onetrust_datasubjects_preferences_bulk_update, onetrust_datasubjects_preferences_bulk_delete, onetrust_consentmanager_v_2_datasubject_profiles_bulk_delete Restrict
Deletes and merges onetrust_requestqueues_bulkdeletes_bulk_update, onetrust_consent_datasubjects_bulk_delete, onetrust_consent_datasubject_purposes_bulk_delete, onetrust_datasubjects_merges_bulk_update, delete_a_onetrust_inventory_inventory_by_id Restrict

The request status writes look small but carry legal weight. onetrust_requestqueue_movestages_bulk_update can move a request to REJECTED or COMPLETE. onetrust_requestqueue_pausedeadlines_bulk_update removes a request's deadline until someone resumes it. onetrust_datasubject_requestqueues_bulk_update adds a comment that can be public rather than internal.

Writing one allow rule for a connector this large

On a connector with hundreds of tools, name what the team reads rather than chase what it must not touch. A restriction is a rule naming which connectors and which individual tools a target may reach. In Elaichi, restriction targets are role or user only; the organization default is the absence of a rule, which means allow everything. So the privacy team needs its own role, held by the people on Emily's team.

Write one allow rule that names the seven read tools. The same rule, or a second allow rule on the role, must also name every other app the team uses as whole connectors. In Elaichi, once a role carries any allow rule, every connector and tool none of its allow rules names is blocked for that role. That is the point: the hundreds of OneTrust tools nobody named stay out of reach. Do not name the onetrust connector whole and some of its tools in one rule. Elaichi refuses that overlap when the rule is created.

Then add block rules on the destructive and status tools in the table. Within a layer, a block always beats an allow. If someone later widens the allow rule to all of OneTrust, the bulk delete and the consent deletes stay restricted. Block rules match a tool's name or its operation, and how block and allow rules match tools explains why. A restriction change takes effect within about two minutes.

Sharing the OneTrust connection and adding the endpoint

A privacy admin connects the onetrust connector once, then shares it with the team at use. A shared connection runs every call on its owner's credential, so the whole team reaches OneTrust through one client credential. Elaichi still records which member made each call.

Pick an owner who will stay. When an owner leaves, offboarding can transfer a shared connection to another active member. Offboarding when an agent holds access covers the rest of that list.

Every organization uses one address: https://api.elaichi.ai/mcp. Each member signs in to it from Claude with their own OAuth grant, the sign-in that ties their Elaichi identity to Claude. Adding Elaichi to Claude walks through the screens, and ChatGPT, Cursor and Microsoft Copilot use the same endpoint. In Elaichi, connected tools are never listed one by one, however few there are. On the consent screen, "Use your connected apps" covers deletes in connected apps as well as reads. The privacy role's rules are what hold Claude to OneTrust reads.

Testing the privacy role before rollout

Sign in from a test account on the privacy role and ask Claude which requests are in progress. The read should run. Then look up one test customer's consent preferences. Then search for "bulkdelete" and "movestages". A restricted tool is left out of the tools Elaichi offers a client and cannot be called. Search still names it, flagged restricted, with no schema.

Then read the audit trail. Elaichi writes one entry for each connected-tool call that reaches execution (a call refused earlier writes none). Each row names the member, the tool, the connection, the outcome and the MCP client, and logs the one path argument that names the object, as the target id, and nothing else about the arguments. For what a privacy reviewer should look for, see the fields an agent audit log needs.

Both sides set rate limits. OneTrust allows 200,000 requests an hour per account and 1,000 a minute per endpoint. It also lists a stricter quota of 300 calls a minute on its consent data subjects endpoint (developer.onetrust.com, checked November 2026). Elaichi allows 120 MCP requests a minute per OAuth grant.

When someone later needs a restricted tool, they file an access request. Only an admin in the Elaichi console approves it, under Governance, then Access requests. A grant approved in the console has no end date, so approve the bulk delete for nobody.

When OneTrust's own MCP options are enough

OneTrust documents two MCP options of its own. The Developer Portal MCP server at https://developer.onetrust.com/mcp serves AI code editors such as Cursor and Windsurf. It works with OneTrust's APIs and documentation and needs no authentication headers (developer.onetrust.com, checked November 2026). If Emily's engineers are writing a DSAR export script, that server fits the job with no second vendor.

OneTrust's September 30, 2026 announcement also describes an MCP Gateway with integrations to applications such as ChatGPT, Claude and Copilot. It was announced as a private preview (onetrust.com, checked November 2026). A team that only needs OneTrust in its AI client should watch that preview.

Elaichi earns its place when the privacy question crosses apps. The request sits in OneTrust, the contract in DocuSign and the evidence in Vanta. Elaichi gives Claude, ChatGPT, Cursor and Microsoft Copilot one endpoint, one set of rules and one audit trail. Vanta for compliance work in Claude and DocuSign held to reads for legal ops cover the next two apps. One person trying a few lookups may not need a control plane yet. The rest is in the 900+ connectors.

FAQ

Frequently asked questions

Does OneTrust have its own MCP server?

OneTrust's Developer Portal documents an MCP server at developer.onetrust.com/mcp for AI code editors such as Cursor and Windsurf, working with its APIs and documentation (checked November 2026). OneTrust's September 30, 2026 announcement also describes an MCP Gateway for applications such as ChatGPT, Claude and Copilot, marked as a private preview.

Which OneTrust scopes does a read-only privacy connection need?

OneTrust offers read scopes per module: DSAR_READ for privacy rights requests, CONSENT_READ for consent and preference data, ASSESSMENT_READ and INVENTORY_READ (checked November 2026). OneTrust lists its Delete Requests endpoint under DSAR_WRITE, not DSAR_READ. Tokens for a credential carry only the scopes assigned to it.

What does onetrust_requestqueues_bulkdeletes_bulk_update do?

In Elaichi's onetrust connector, this tool permanently deletes OneTrust DSAR requests in bulk, with their subtasks, attachments and Results Summaries. It can select requests by date range, stage, country or request type, deletes up to 50 per call and reports the rest as pending. Elaichi tiers it destructive, so a privacy role should keep it restricted.

How quickly does a new OneTrust restriction take effect in Elaichi?

A restriction change in Elaichi takes effect within about two minutes, so test a new rule after that window. Revoking a share is faster: in Elaichi, revoking a share or disconnecting an account takes effect on the caller's next call.

Put AI to work in your own apps

14 days of Black, no credit card. Start with one app and one team.

When the trial ends, nothing is deleted. You move to the free Personal plan, and connections, roles and the audit log stay where they are.

Add one link to your AI app

The same link for everyone on your team.

https://api.elaichi.ai/mcp
  • ChatGPT
  • Claude
  • Cursor
  • Microsoft Copilot