Describe the Bug
When sending a complex prompt to the Perplexity API, the request is blocked or fails, even though it should be processed normally. The system doesn’t provide a clear error message and unexpectedly interrupts the workflow.
Expected Behavior
The API should accept the prompt, process it, and return a valid, non-blocking response.
I’m only including a technical analysis prompt in our tool to improve our service. Otherwise, the tool will be useless.
Actual Behavior
The API blocks the request or returns an error indicating that the input is not allowed, even though the input is neutral and compatible with normal use.
Steps to Reproduce
Prepare a prompt that requires code analysis or technical generation (e.g., chatbot automation code).
Send the prompt to the Perplexity API.
Observe whether the request is blocked or fails unexpectedly.
API Request and Response (if applicable)
Request: Prompt containing a request for technical code analysis.
Response: Error message or blocked request (exact message varies).
Environment
API Version: Latest Perplexity API
SDK: N/A (tested directly via API)
Operating System: Tested on Windows 10 and Linux
Logs or Screenshots (if applicable)
Screenshot of the blocked request or API error message.
Additional Context
The issue appears to be related to certain keywords in the prompt (addons, “create”, “make”, “code”).
Rewriting the prompt with neutral language may prevent the block.
The user needs a solution to allow technical code analysis prompts without being blocked.
