[Bug] Repetitive and unhelpful responses when executing tools after an initially correct query

[Chikedor]

Description:

After making an initial query that was answered correctly (in this case, asking "What is MiniPerplx?"), the AI started generating repetitive and unhelpful responses when asked for a description for a comparison table. In these responses, it repeatedly stated that it was executing the thinking_canvas tool but failed to generate any meaningful result or explain the reason for the failure.

This behavior impacts the user experience, especially since the flow initially worked as expected.

---

Expected Behavior:

• Provide a useful and relevant response to the request for a description of a comparison table.
• If unable to process the request, provide a clear and concise explanation (e.g., "Unable to generate a response at this time due to X reason").
• Avoid repetitive and redundant responses that do not add value.

---

Current Behavior:

1. The AI correctly answered the initial query: "What is MiniPerplx?"
2. Upon requesting a description for a comparison table, the AI began generating repetitive responses stating it was executing the thinking_canvas tool without providing any useful output.

---

Steps to Reproduce:

1. Ask the AI: "What is MiniPerplx?"
2. Follow up with: "Provide a short description for a comparison table."
3. Observe how the AI responds with repetitive and unhelpful answers.

---

Evidence:

Attached screenshot:

---

Environment:

• Browser or Platform: Brave on iOS.
• Date and Time of Incident: 18:00 GMT +1 aprox.

---

Additional Notes:

• This behavior may indicate a technical issue in the integration or handling of the thinking_canvas tool.
• It might be helpful to review state handling between consecutive queries since the flow failed after the initial correct response.
• Consider adding clearer error messages if the tool cannot process a request.

---