Skip to content

Conversation

@sidhpurwala-huzaifa
Copy link
Owner

Summary

  • strengthen dangerous capability detection by analysing schema guardrails and parameter metadata
  • improve injection fuzzing heuristics to test all string parameters and record execution indicators
  • add targeted unit tests covering the enhanced detection logic

Testing

  • pytest

https://chatgpt.com/codex/tasks/task_e_6904afab63a88332adc43cbfbb1b88fb

)

if tool_reasons:
risky_tools.append(
Copy link
Contributor

@ccronca ccronca Nov 4, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This check now returns a simplified structure instead of the full tool object (for example, tool.metadata is no longer included). Was this change intentional?

We could preserve the previous behavior with something like:

risky_tools.append({
      **tool, 
      "reasons": tool_reasons, 
  })

"tools/call",
{"name": name, "arguments": {param_name: payload}},
)
except Exception:
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Maybe log something here, otherwise, tracking issues could be tricky

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants