Reading content of mcp tool calls

Support Bug Reports
,
1

Where does the bug appear (feature/product)? MCP & tools

Describe the Bug In previous versions of Cursor I used to be able to read/expand the content of MCP tool calls. This was very useful to quickly audit Supabase MCP tool calls before pressing allow, especially when running execute_sql or other Supabase tools. Now I can only see the tool name and press allow/deny, so I am not able to review the command being executed to determine if I should allow it or not. Previously the command was shown in a code block and included the Cursor tool call guardrails.

Steps to Reproduce

  1. Connect the Supabase MCP server (I have mine set to read only)
  2. Trigger a tool call, e.g. execute_sql
  3. Observe the allow/deny prompt: only the tool name is shown, with no way to expand and view the actual command/SQL being run

Expected Behavior The tool call prompt should show the command or SQL being run in a code block before I allow or deny it, the way it did in previous versions.

Screenshots / Screen Recordings

image
image500×227 4.94 KB

Operating System Windows 11

Version Information
Version: 3.12.17 (user setup)
VS Code Extension API: 1.128.0
Date: 2026-07-17T02:53:53.006Z
Layout: IDE
Build Type: Stable
Release Track: Default
Electron: 40.10.3
Chromium: 144.0.7559.236
xterm.js: 6.1.0-beta.256
OS: Windows_NT x64 10.0.26200

For AI issues: which model did you use? N/A, this is a UI/display issue, not model specific

For AI issues: Request ID N/A

Additional Information I looked through the settings and could not see anywhere to re-enable this option, although I have the conversation density set to “Detailed,” which is described as “Choose how much detail Agent tool calls show in the conversation.”

I hope this is temporary and this feature is being improved and will show up again in new updates. Appreciate letting us know if this is in the pipeline, or I should set up my own tooling to ask the agent to write tool calls to a file before executing them so I can audit.

I currently have Supabase MCP set as read only, so this is mildly annoying. However, I cannot envisage setting up the MCP as read/write without having the ability to review tool call contents, which can be potentially destructive or alter my databases.

Does this stop you from using Cursor? Sometimes - I can sometimes use Cursor when mcp is setup to read-only and hoping it is running correct commands checking correct tables…

Edit: updated this post to follow the bug report template per the mod bot’s request above.

2

Hi there!

We detected that this may be a bug report, so we’ve moved your post to the Bug Reports category.

To help us investigate and fix this faster, could you edit your original post to include the details from the template below?

Bug Report Template - Click to expand

Where does the bug appear (feature/product)?

  • Editor, Tab & Chat (autocomplete, Composer, in-editor agent)
  • Terminal & commands
  • Models, pricing & API keys (availability, Auto/Max, BYOK/Bedrock)
  • MCP & tools
  • Cloud Agents & Automations (cursor.com/agents, scheduled/event)
  • BugBot & Code Review
  • Cursor CLI
  • Cursor Mobile
  • Remote (SSH / Dev Containers / WSL)
  • Account, billing & login
  • Something else…

Describe the Bug
A clear and concise description of what the bug is.

Steps to Reproduce
How can you reproduce this bug? We have a much better chance at fixing issues if we can reproduce them!

  1. …
  2. …
  3. …

Expected Behavior
What is meant to happen here that isn’t working correctly?

Screenshots / Screen Recordings
If applicable, attach images or videos (.jpg, .png, .gif, .mp4, .mov)

Operating System

  • Windows 10/11
  • MacOS
  • Linux

Version Information

  • For Cursor IDE: Menu → About Cursor → Copy
  • For Cursor CLI: Run agent about in your terminal
IDE:
Version: 2.xx.x
VSCode Version: 1.105.1
Commit: ......

CLI:
CLI Version 2026.01.17-d239e66

For AI issues: which model did you use?
Model name (e.g., Sonnet 4, Tab…)

For AI issues: add Request ID with privacy disabled
Request ID: f9a7046a-279b-47e5-ab48-6e8dc12daba1
For Background Agent issues, also post the ID: bc-…

Additional Information
Add any other context about the problem here.

Does this stop you from using Cursor?

  • Yes - Cursor is unusable
  • Sometimes - I can sometimes use Cursor
  • No - Cursor works, but with this issue

The more details you provide, the easier it is for us to reproduce and fix the issue. Thanks!

7

Hey there,

You’re right, and this isn’t something on your end. A recent change to how tool calls render stopped surfacing MCP arguments, and on the approval prompt they aren’t shown at all right now, so there’s currently no in-UI way to review the input (like the execute_sql SQL) before you allow or deny. I’ve reproduced it on my side and reported it to the team.

On the setting: the Conversation Density “Detailed” option won’t bring it back. It only changes how completed tool calls are grouped in the transcript, not what the approval prompt shows, so that one is a bit of a red herring here.

Until it’s fixed, your instinct about writing the call out first is the safest approach:

  1. Keep risky Supabase tools (like execute_sql) off Always Run, so you always get the Skip option instead of an auto-approve.
  2. Ask the agent to state the exact SQL and arguments in chat before it calls the tool, so you can review them and then approve.

I’m going to fold this into the main thread where we’re already tracking the same issue, so all the reports stay together and you’ll see updates in one place: Can not see call args in mcp tool calls. I’ll post there as soon as there’s progress.