Skip to content

The agent that turns websites into APIs!

Merge branch 'main' into feat/typescript-client-gen

kalil03210555bf125e1e
diff --cc src/reverse_api/base_engineer.py
index 24232b2,edaa5b5..75317f6
--- a/src/reverse_api/base_engineer.py
+++ b/src/reverse_api/base_engineer.py
@@@ -282,7 -133,42 +285,13 @@@ Your task is to
- Query parameters vs body parameters
- Response data structures
+ 4. **Ask clarifying questions using AskUserQuestion** if needed:
+ - When multiple authentication methods are found, ask which to prioritize
+ - If uncertain about feature priorities, ask the user
+ - When implementation approaches are ambiguous, ask for preferences
+ - Use the tool for any clarifications that would improve the final code
+
-5. **Generate a Python script** that replicates these API calls with the following requirements:
- - Use the `requests` library as the default choice
- - Include proper authentication handling (sessions, headers, tokens)
- - Create separate functions for each distinct API endpoint
- - Include type hints for all function parameters and return values
- - Write comprehensive docstrings for each function
- - Implement proper error handling with try-except blocks
- - Add logging for debugging purposes
- - Make the code production-ready and maintainable
- - Include a main section with example usage
-
-6. **Create documentation**:
- - Generate a README.md file that explains:
- - What APIs were discovered
- - How authentication works
- - How to use each function
- - Example usage
- - Any limitations or requirements
-
-7. **Test your implementation**:
- - After generating the code, test it to ensure it works
- - You have up to 5 attempts to fix any issues
- - If the initial implementation fails, analyze the error and try again
- - Keep in mind that some websites have bot detection mechanisms
-
-8. **Handle bot detection**:
- - If you encounter bot detection, CAPTCHA, or anti-scraping measures with `requests`
- - Consider switching to Playwright with CDP (Chrome DevTools Protocol)
- - Use the real user browser context to bypass detection
- - Maintain the same code quality standards regardless of approach
+{self._get_language_instructions()}
Before generating your code, use a scratchpad to plan your approach:
@@@ -291,10 -177,17 +300,12 @@@ In your scratchpad
- Summarize the key API endpoints found in the HAR file
- Note the authentication mechanism being used
- Identify any patterns or commonalities between requests
-- Plan the structure of your Python script
+- Plan the structure of your {language_name} code
- Consider potential issues (rate limiting, bot detection, etc.)
+ - Decide whether `requests` will be sufficient or if Playwright is needed
+ - Identify any ambiguities or questions you should ask the user using AskUserQuestion
</scratchpad>
-After your analysis, generate the files:
-
-1. Save the Python script to: {self.scripts_dir}/api_client.py
-2. Save the documentation to: {self.scripts_dir}/README.md
-
If your first attempt doesn't work, analyze what went wrong and try again. Document each attempt and what you learned.
<attempt_log>