CHANGELOG.md

[0.4.0] - 2025-09-12

New & Enhanced Features

• Google Search API integration
• Chatbot development example
• LLM-Graphbit-Playwright Browser Automation Agent
• Text splitter module
• DeepSeek provider support
• Perplexity provider support
• Complete MkDocs docs site (Material theme)
• Docs site (initial)
• Rust Core & Python binding: Agentic workflow withdep-batching & parent preamble
• Node.js binding
• Tool calling support
• Makefile: cross-platform test infra + install; Rust/Python test targets
• System prompt added
• LLM configuration per node

Bug Fixes & Stability

• Resolve hook failures & improve security compliance
• Update black & dependency checks
• macOS fallback for sched_getaffinity
• Codebase formatted to pass pre-commit
• Benchmark CPU affinity logic cross-platform (macOS support, safe fallback)
• Python tests: add/improve LLM, executor, doc loader tests
• Anthropic LLM config issue fixed
• Pre-commit hook error fixed
• Perplexity prompt error fixed
• Makefile refactor fix
• LlamaIndex missing dependency fixed
• Sentence splitter issue fixed
• Character splitter infinite loop fixed
• Executor dict→JSON conversion fix
• Document loader support in Python binding
• Fixed Rust & Python test bugs

Refactors & DX

• Removed docs site & related files/code
• graphbit init call auto by default
• Remove unnecessary packages from pyproject.toml
• Refactor: import graphbit in benchmark
• API reference folder refactor
• Node.js bindings & related code refactor
• Remove warning for Python binding
• Version management: caching + improved reporting
• pyproject.toml cleanup
• Remove pyproject.toml and update Cargo.toml for Python consistency
• Refactor benchmark scripts
• Getting-started folder refactor

Documentation & Guides

• Integrations: MongoDB , Pinecone , Qdrant , PGVector , ChromaDB , FAISS , Milvus , Weaviate , IBM Db2 , Elasticsearch , Azure , GCP , AWS boto3
• Embeddings docs updated ; llm-providers.md ; concepts/embeddings/async-vs-sync ; dynamics-graph ; monitoring ; agents ; document loaders ; performance ; reliability ; memory management ; tool-calling docs ; node types with llm_config
• README updates: root & examples ; architecture diagram & doc validation
• Benchmarks & reports: comparison summary , updated reports , benchmark README
• Docs site content sync (2025-09-09) ; docs/examples folder ; CHANGELOG update ; about file ; validation docs ; remove Hugging Face refs ; license added

Tests

• LLM clients coverage; align workflow connect tests
• Doc processing, text splitting, embedding tests
• Basic unit testing
• Edge cases: client & configuration failure tests
• Python: import/init/security/workflow failure tests
• LLM/agent/workflow tests; validation & integration refactor
• Types/validation/error unit tests
• Doc loader/embedding/text splitter tests
• Python binding tests; Rust helper funcs
• Rust unit tests for agent, concurrency & graph
• Document loader & serialization tests
• Rust types & error tests
• Integration tests: doc load, workflow & error

CI/CD & Workflows

• Auto version update on release & changelog generator
• Validation workflow & runner scripts
• Comprehensive testing workflow & runner
• Build workflow w/ artifact generation & verification
• Deployment action workflow
• Workflow helper files test suite
• CI/CD backup copies workflow
• Pipeline orchestrator workflow
• Sync docs to graphbit-docs repo
• Workflow refactor & updates

Security & Meta

• Add SECURITY.md

[0.3.0-alpha] – 2025-07-16

• Comprehensive Python tests added
• Rust integration test coverage improved
• Improved benchmark runner
• Benchmark run consistency improved
• Centralized control of benchmark run counts
• Centralized run configuration committed
• Explicit flags/config for run counts
• Explicit run control documented
• Dockerization support for benchmark
• Production volume mount paths refined
• Tarpaulin coverage added for Rust
• Tarpaulin configuration integrated
• Benchmark documentation updated
• Root README updated
• Contributing guidelines updated

[0.2.0-alpha] – 2025-06-28

• Ollama Python support added
• LangGraph integrated into benchmark framework
• CrewAI benchmark scenarios optimized for performance and reliability
• Performance optimizations
• Benchmark and Python binding refactors
• New integration tests added
• Python integration tests expanded
• Fixed Python integration tests for GitHub Actions
• Python examples for GraphBit added
• Benchmark evaluation updated
• Example code updated
• Python documentation updated
• Pre-commit issues resolved
• Pre-test commit fixed for all files
• Makefile fixes
• Root README updated
• GitHub Actions workflow removed

[0.1.0] - 2025-06-11

• Initial GraphBit release: declarative agentic workflow framework
• Modular architecture with core modules (agents, llm, graph, validation, workflow, types, errors)
• Multi-LLM support: OpenAI GPT, Anthropic Claude, Ollama, extensible providers
• Graph-based workflows (DAG), dependency management, topological execution
• Node types: agent, transform, conditional
• Parallel execution with configurable concurrency
• Async/await support throughout
• Full Python API via PyO3 with async support
• Strong typing and validation
• JSON schema validation for LLM outputs
• UUID identifiers for all components
• Intelligent dependency resolution
• Error handling: retries with backoff, comprehensive errors, failure recovery
• Usage tracking: tokens, cost estimation, performance metrics
• CLI (graphbit): init, validate, run, config, debug/verbose
• JSON workflow configs with env var support and custom files
• Integration examples: FastAPI, Django, Jupyter
• Documentation and examples: README, Ollama guide, testing/benchmarking, Python and Rust API docs
• Testing: unit/integration, benchmarking suite, mock LLM providers, CI/CD
• Dependencies and MSRV: tokio, serde, anyhow/thiserror, uuid, petgraph, reqwest, clap, pyo3, chrono; Rust 1.70+