Skip to main content

Connecting Requidex MCP in Claude and Cowork

How to add the Requidex MCP server as a remote custom connector in Claude and Cowork

Written by Sam

What this connector does

The Requidex MCP connector lets Claude and Cowork call read-only Requidex Open API tools from a conversation and search Requidex Intercom help guides. It can list available Open API operations, query permitted Requidex data such as projects, assignments, timesheets, invoices, workers, trades, companies, and related records, and fetch guide content for product knowledge.

Access is based on the Requidex user who completes the OAuth sign-in. Only Requidex company admin accounts can connect, and the connector uses that user's company scope.


Before you start

  • Use the MCP server URL: https://api.requidex.com/mcp

  • Sign in with an active, verified Requidex company admin account.

  • Use a Claude plan and workspace that allows custom connectors.

  • For Team and Enterprise Claude workspaces, an Owner or Primary Owner must add the custom connector before members can connect it.


Add the connector

  1. Open Claude connector settings. Team and Enterprise owners use Organization settings > Connectors. Individual Pro or Max users use Customize > Connectors.

  2. Choose Add, then choose Custom and Web.

  3. Enter https://api.requidex.com/mcp as the remote MCP server URL.

  4. Leave advanced OAuth credentials blank unless Requidex support has given you specific client credentials.

  5. Save the connector.


Connect your account

  1. Open Customize > Connectors.

  2. Find the Requidex custom connector.

  3. Select Connect.

  4. Complete the Requidex sign-in and any MFA prompt in the browser.

  5. Return to Claude or Cowork and enable the connector for the conversation from the connectors or tools menu.


Using the tools

Ask Claude to use the Requidex connector when you need current operational data from Requidex. The connector is read-only, so it can retrieve data but cannot create, update, approve, cancel, or delete records.

When you ask by company, project, agency, site, or trade name, the connector can use lookup tools to resolve those names to ids before calling filtered Open API tools.

For spend, total, count, ranking, top-N, or grouped analysis questions, the connector can use the relevant summary tool for invoices, assignments, timesheets, requisitions, expenses, or credit notes instead of fetching raw rows.

For list tools that support field selection, use the availableFields returned by list_open_api_operations or the fields shown in each tool description before setting the fields parameter.

For list tools that expose dateField, from and to apply to that selected date field. Use dateField when the user asks for start dates, end dates, week-ending dates, approval dates, confirmation dates, due dates, processed dates, or expense dates instead of creation dates.

Paginated tool calls default to 50 records, return at most 200 records per page, and can page through up to the first 10000 matching records. They include the total count so you can narrow broad requests with filters. For count-only or existence checks on list tools, set limit=1 and read meta.total.

When the question is about how Requidex works, use the search tool to find matching Requidex help guides and the fetch tool to read the full guide before answering.

If the model has multiple data sources available, name Requidex in the prompt. For example: Use the Requidex connector to list active projects for this company.


Advanced filtering and aggregation

List and summary tools support multi-company filtering, and many tools support multi-ID filtering for documented entity filters such as project, site, trade, worker, assignment, timesheet, requisition, and agency. Pass comma-separated ids where the tool description allows it. For example, company='id1,id2,id3' returns data for all three companies instead of requiring three separate calls.

List tools that support the fields parameter accept dot-path nested fields as well as top-level fields. For example, fields='worker.id,values.charge.total,values.charge.wdBasic' returns only those nested timesheet fields instead of the full values object.

List tools that support dateField apply from and to to the selected date field. For example, use dateField='weekEnding' for timesheet week filters, dateField='dateAccepted' for requisition agency acceptance filters, and dateField='issueDate' for invoice issue-date billing periods.

All summary tools support multi-dimensional aggregation through the groupBy parameter and grouped result sorting through sortBy, sortOrder, and limit. Use groupBy to break down results by one or more dimensions such as company, trade, agency, purchaseOrder, status, category, or month. For example, groupBy='company,trade' returns spend per trade per company in one call instead of N×M separate calls, and groupBy='agency' with sortBy='finalTotal' and limit=5 returns the top five agencies by invoice value. Allowed values vary by endpoint and are documented in each summary tool description.

Timesheet summaries can group by worker, trade, and purchaseOrder, return distinctWorkers, and include charge and pay splits for weekday basic, weekday overtime, nightwork, weekend, and bank holiday values.

The recruitment metrics list tool can return server-side aggregates by setting metric='timeToFill' or metric='workersOverHours'. These modes support agency and period filters, groupBy values such as agency, project, month, and week, compound groupBy values such as company,agency, the IQR outlier rule for time-to-fill, and a configurable weekly hours threshold for workers-over-hours.

Combining multi-ID filtering with groupBy enables powerful cross-entity analysis. For example, summarize_assignments with company='id1,id2,id3' and groupBy='trade,month' returns assignment spend broken down by trade and month across three companies in a single API call.


Troubleshooting

  • If connection fails immediately, confirm the server URL is exactly https://api.requidex.com/mcp.

  • If authentication fails, disconnect and reconnect the connector so Claude starts a fresh OAuth flow.

  • If you see no data, confirm your Requidex account is active, verified, and a company admin with access to the relevant company.

  • If a Team or Enterprise member cannot find the connector, ask a workspace Owner to add it at organization level first.

  • Remote connectors are reached from the AI client's cloud infrastructure, so private networks or firewall rules can block access even if the URL works from your browser.


For agencies

This connector signs in with the Requidex OAuth flow, which is available to hiring-company admin accounts. Agency and desk accounts cannot connect through this OAuth flow.

  • If you work for an agency and need programmatic access to your own data, ask your administrator about agency-scoped Open API keys instead of this connector.

  • Agency-scoped access is read-only and limited to the agencies you are connected to.

  • Some company-level endpoints, such as projects, trades, sites, and purchase orders, are not available to agency-scoped access.

Did this answer your question?