Conversation
- Add detailed configuration section covering zero configuration setup - Document environment variables including OPENAI_API_KEY, VECTORLESS_MODEL, VECTORLESS_ENDPOINT, and VECTORLESS_WORKSPACE - Provide Python and Rust code examples for different configuration methods - Explain configuration priority hierarchy from defaults to constructor parameters - Update examples reference to include more usage patterns feat(engine): enhance configuration with environment variable support - Implement environment variable overrides for API key, model, endpoint, and workspace - Add configuration priority system with 5 levels from defaults to constructor parameters - Support zero configuration through OPENAI_API_KEY environment variable - Add config_path parameter for advanced configuration file usage - Update Python and Rust engines with new configuration options refactor(examples): reorganize examples by language with new configuration patterns - Move Python examples to examples/python/ directory - Move Rust examples to examples/rust/ directory - Add basic, advanced, and custom_config examples for both languages - Demonstrate zero configuration, full config file, and custom provider setups - Include examples for various LLM providers like DeepSeek, Azure OpenAI, and local LLMs docs(python): update Python engine documentation with configuration examples - Add configuration priority documentation - Include zero configuration and custom model usage examples - Document constructor parameters workspace, config_path, api_key, model, and endpoint - Show configuration hierarchy from defaults to constructor parameters docs(rust): enhance Rust builder documentation with configuration guide - Document configuration priority system with 5 levels - Add environment variable documentation table - Provide zero configuration, custom model, and full config file examples - Update builder methods to reflect new configuration approach build(rust): register Rust examples in Cargo.toml - Add all Rust example binaries to Cargo.toml manifest - Include examples for basic, advanced, custom_config, and other use cases - Register batch processing, CLI tool, content aggregation examples - Add storage and strategy pattern examples
…ories - Remove old advanced.py and custom_config.py files - Add new example structure with dedicated directories for basic, advanced, and custom_config examples - Each example now has its own README.md, main.py, and pyproject.toml - Update basic example to use from_content instead of from_text - Add proper configuration examples and documentation fix(python): improve IndexContext and Engine class naming - Rename from_text to from_content for better semantic meaning - Add explicit class names for Python bindings (IndexContext, QueryResult, DocumentInfo, Engine) - Update error handling with proper constructor - Fix module name from _vectorless to vectorless refactor(rust): update document format handling and builder logic - Remove Text format support temporarily - Set default format to Markdown instead of Text - Fix EngineBuilder to properly handle API key and model precedence - Update with_openai to not override existing model configuration - Improve documentation for builder methods
- Implement comprehensive API key and configuration propagation across retrieval and summary components in EngineBuilder - Enable default summary generation by setting generate_summaries to true in IndexOptions - Add extensive debugging output throughout the indexing and retrieval pipeline for better observability - Refactor LLM pilot to support candidate title matching and improved response parsing with flexible JSON formats - Update prompts to enforce strict JSON-only responses with explicit structure requirements - Modify beam search to integrate pilot guidance at start and implement enhanced scoring normalization The changes improve the robustness of LLM-based document navigation and provide better visibility into the decision-making process through comprehensive logging while maintaining backward compatibility.
…ates - Add LocateTop3Prompt for identifying top-3 relevant nodes from TOC - Implement query decomposition in AnalyzeStage for complex queries - Add Top3Candidate struct for handling LLM response parsing - Enhance prompt templates with strict JSON formatting requirements - Remove redundant default_decision test - Update SearchStage to handle decomposed sub-queries with multi-turn processing - Add decomposition field to PipelineContext to store sub-query results - Configure complexity thresholds for triggering decomposition BREAKING CHANGE: Updated prompt formats require strict JSON responses without markdown code blocks.
Add new LLM-first search capability that attempts to directly locate relevant nodes using the table of contents before falling back to tree traversal algorithms. The search stage now accepts an LLM client and implements direct TOC-based node location with structured JSON responses containing top-3 most relevant entries. The feature includes: - New `with_llm_client()` method to configure LLM-based search - TOC flattening utility for LLM consumption with numbered entries - Structured JSON prompting for precise node selection - Proper fallback to beam/greedy search when LLM fails - Metrics tracking for LLM calls and search performance refactor(indexer): change default summary strategy to full Change the default summary strategy from selective to full generation for improved content indexing quality. debug: add debug logging throughout indexing and retrieval Add comprehensive debug logging to track indexing flow, pipeline options building, summary evaluation, and search operations including pilot interventions for better debugging visibility.
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
No description provided.