Add haiku explanation type with flexible prompt composition #6
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…mpt composition - Add haiku-specific test cases covering simple to complex algorithms - Implement structure-aware methods in Prompt class for better code reuse - Update prompt_advisor.py to handle nested audience_levels in explanation types - Add haiku-specific evaluation criteria in claude_reviewer.py - Update documentation to explain flexible prompt composition system - Support explanation-type overrides for audience guidance (enables haiku to bypass assembly-specific instructions) 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <[email protected]>
- Remove duplication between get_audience_metadata methods - implement instance method in terms of class method - Remove overly complex get_all_audience_locations method, add comment about future needs - Remove hardcoded haiku-specific evaluation criteria - use same criteria for all explanation types - Remove magic string detection for haiku targeting - explanation types should be explicitly specified - Simplify prompt_advisor.py to directly check both audience locations without complex abstractions - Fix test to check audience information in user prompt where it actually appears - Use .values() instead of .items() when we don't need the keys 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <[email protected]>
There was a problem hiding this comment.
Pull Request Overview
Adds a haiku explanation type to the Claude assembly explanation service with a flexible prompt composition system that allows explanation types to override audience-specific guidance. This enables poetic explanations that bypass technical assembly guidance while maintaining audience differentiation for assembly explanations.
- Added
ExplanationType.HAIKUenum value and haiku-specific prompt configuration - Implemented explanation-type audience overrides in prompt system with structure-aware class methods
- Enhanced testing framework with haiku test cases and Claude evaluation criteria
Reviewed Changes
Copilot reviewed 11 out of 11 changed files in this pull request and generated 2 comments.
Show a summary per file
| File | Description |
|---|---|
| prompt_testing/test_cases/haiku_tests.yaml | Comprehensive haiku test cases covering various algorithm complexities |
| prompt_testing/evaluation/prompt_advisor.py | Enhanced to handle nested explanation-specific audience configurations |
| prompt_testing/evaluation/claude_reviewer.py | Added haiku-specific evaluation criteria |
| claude_explain.md | Updated documentation with flexible prompt system details |
| app/test_explain.py | Fixed test to check audience guidance in user message instead of system prompt |
| app/prompt.yaml | Added haiku configuration with audience overrides and restructured prompts |
| app/prompt.py | Added structure-aware class methods for prompt composition |
| app/main.py | Added configurable logging via LOG_LEVEL environment variable |
| app/explanation_types.py | Added HAIKU enum value |
| app/explain.py | Added debug logging for actual prompts sent to Claude |
| app/config.py | Added log_level configuration setting |
Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.
- Replace module-level resource initialization with lifespan context manager - Create configure_logging() function for proper logging setup - Move shared resources (anthropic_client, prompt, settings) to app.state - Update endpoints to access resources from app.state via Request object - Eliminate multiple get_settings() calls at module import time - Use proper FastAPI 2024 best practices for resource management - Maintain backward compatibility and all tests pass This fixes the issue where settings were locked at import time and makes testing much easier since settings can now be mocked before app creation. 🤖 Generated with [Claude Code](https://claude.ai/code) Co-Authored-By: Claude <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary
Adds haiku explanation type and flexible prompt composition system that allows explanation types to override audience-specific guidance.
Changes Made
Core Implementation
ExplanationType.HAIKUenum valueTesting & Evaluation
prompt_testing/test_cases/haiku_tests.yamlwith test scenariosclaude_reviewer.pywith haiku-specific evaluation criteriaprompt_advisor.pyto handle nested audience configurationsCode Quality & Reuse
Promptclass:get_audience_metadata_from_dict()- Works with raw prompt dictionarieshas_audience_override()- Checks for explanation-specific overridesget_all_audience_locations()- Finds all audience guidance locationsDebug Support
LOG_LEVELenvironment variableDocumentation
claude_explain.mdwith flexible prompt system documentationTest Plan