# Tradly MCP: Scoped Agent Access, Audit Logs, and Agentic Workflows **URL:** \ **Category:** Product Updates\ **Created:** [18 September 2026 06:01 UTC](http://community.tradly.app/t/tradly-mcp-scoped-agent-access-audit-logs-and-agentic-workflows/4949 "2026-09-18T06:01:29Z")\ **Posts on this page:** 1\ **Showing post:** 1
**Author:** ![jkbaseer](http://community.tradly.app/user_avatar/community.tradly.app/jkbaseer/32/1296_2.png) [@jkbaseer](http://community.tradly.app/u/jkbaseer)\ **Post date:** [18 September 2026 06:01 UTC](http://community.tradly.app/t/tradly-mcp-scoped-agent-access-audit-logs-and-agentic-workflows/4949/1 "2026-09-18T06:01:29Z")
We’ve been investing heavily in **MCP at Tradly.app since August 2025** , as the ecosystem around AI agents, tools, authentication, and permissions has continued to mature. MCP itself was still relatively new when we started. Standards were evolving, documentation was changing quickly, and authentication implementations differed across platforms. We spent a significant amount of time working through OAuth requirements, following discussions in the OpenAI developer community, and adapting our implementation as the ecosystem developed. At the same time, MCP usage was still early among many of our customers, so we deliberately kept our implementation in the background while the standards matured. That changed with the launch of **Tradly AI Studio — aistudio.tradly.app**. As customers started building custom agents and workflows, MCP became much more important. Agents need access to real business capabilities: analytics, content, commerce, marketing, customer data, and operational tools. But giving an agent access to tools also creates an important question: **What exactly should that agent be allowed to do?** ## Introducing scoped MCP access ![MCP Scopes](http://community.tradly.app/uploads/default/original/2X/0/0436b6d843d4f7a85be9c46046765e7f38966d79.png) We have now extended Tradly’s permission model to MCP connections and AI agents. Following the same **role-based access control principles** used across Tradly, businesses can define which capabilities an authorised agent is allowed to access. Instead of giving an agent unrestricted access to an entire account, permissions can be scoped to the tools and actions that are relevant to its workflow. We have already introduced several MCP scopes, with **150–200 additional tool calls** being developed and tested internally across the Tradly platform. ## One conversation, multiple business workflows The example below shows **mcp.tradly.app connected to ChatGPT** , with the agent interacting with Tradly’s analytics tools through a conversation. ![Tradly MCP in action](http://community.tradly.app/uploads/default/original/2X/9/9762bb54e230b50fc2509619d0cbb5f32924b7f8.png) But analytics is only the beginning. Imagine asking: “What happened to sign-ups this week?” Then: “Create a CMS article explaining the trend.” And finally: “Send a marketing campaign to the relevant customer segment.” All from the same conversation, while the underlying actions happen securely inside the same Tradly data platform. For marketers and small business owners, this means less switching between dashboards, tools, and interfaces. The AI agent can become the interface, while Tradly remains the system executing the underlying workflows. ## Three important capabilities in Tradly MCP ![MCP connections](http://community.tradly.app/uploads/default/original/2X/e/efe62e3934a9674390d00d490292be0e01c9024f.png) **1. Agent permissions can be changed after authorisation** Connecting an agent does not mean giving it permanent unrestricted access. Administrators can later modify or restrict the scopes available to an authorised agent. **2. Additional MCP connections can be managed centrally** New agent and MCP connections can be added and managed from the same environment, making it easier for businesses to control how AI systems interact with their Tradly account. **3. Agent actions will be included in audit logs** ![image](http://community.tradly.app/uploads/default/original/2X/c/cf65298ec02807c28abaa12eb6bbb62ae0063df1.jpeg) Tradly already records audit trails for actions performed by human users. We are extending the same approach to AI agents. This means businesses can understand **which agent performed an action, what it accessed, and what happened inside the platform**. As agents become capable of performing more operational tasks, this level of visibility becomes increasingly important. ## Building toward agentic commerce For us, building MCP has been as much a learning process as a product development effort. The infrastructure around agents, authentication, permissions, tools, and interoperability continues to evolve quickly. We’re continuing to experiment with how these technologies can make commerce and business operations significantly easier. The direction is becoming clearer: businesses should not need to navigate five different interfaces just to understand their data, create content, launch a campaign, and take action. Agents can increasingly coordinate those workflows. Our job at Tradly is to make sure Agents can do it with the right **tools, permissions, data access, and auditability**. More updates on what we’re building around **agentic commerce** are coming soon. --- _[View the full topic](http://community.tradly.app/t/tradly-mcp-scoped-agent-access-audit-logs-and-agentic-workflows/4949)._