✅ v0.2.0

🧠 New Features: Chunking and Similarity Modules

This is a major milestone toward intelligent data handling.

• chunker.ts: Text chunking with precision split control
• markdown-splitter.ts: Intelligent Markdown-aware splitting
• cosine-similarity.ts: Semantic similarity calculator
• Test suite updated for full coverage of these features
• tsconfig.json modified to explicitly include test paths

📝 Changes

+ New: `chunker`, `cosine-similarity`, `markdown-splitter` in `src/core/`
+ New tests: `chunker.test.ts`, `md-splitter.test.ts`
~ Updated exports via `src/core/index.ts`
~ `tsconfig.json` includes `src/tests`

v0.1.5

➕ Dependency Additions

• Installed new packages to support future modules
• Lockfile (bun.lockb) updated

No code changes beyond dependency additions.

v0.1.4

📦 Refactor: Project layout

• Moved all tests to a top-level src/tests folder
• Cleaned up path resolution in core modules
• Minor internal restructuring for consistency

📝 Changes

→ Moved test files to `src/tests/`
~ Modified: `ollama-service.ts`, `ufetch.ts`

v0.1.3

🧹 Minor fixes and test correction

• Test file corrections for new file structure
• Regression fixes for OpenAI and Ollama tests

📝 Changes

* Fixed: Test failures in `openai-service.test.ts`, `ollama-service.test.ts`
* Updated: `app.md`

v0.1.2

✨ What’s new

• Clean-up prior to restructuring package boundaries
• Introduced prompt types and structure
• Added .npmrc to prepare for registry configuration

📝 Changes

+ Added: `src/types/prompts.ts`, `.npmrc`
+ Updated: `prompts.ts`, `package.json`, `bun.lockb`
+ Removed: Legacy files in preparation for modularisation

v0.1.1 – Initial Alpha Release

This marks the first public release of @jasonnathan/llm-core, establishing the foundational structure and functionality of the library.

Highlights

• Initial Core Architecture:

  • Introduced modular design with reusable LLM integration utilities.
  • Established baseline for pipeline orchestration, OpenAI/Ollama clients, decorators, and runtime helpers.
  • Included semantic utilities like similarity-service, classification-service, and ml-service.

• Testing & Stability:

  • Added Jest-based test suite with mock utilities for OpenAI and Ollama.
  • Verified pipeline and logger behaviour via unit tests.
  • GitHub Actions workflow set up for publish and CI.

Module Introductions

• src/core/ollama-service.ts – Typed local LLM client with validation and structured output.
• src/core/openai-service.ts – Lightweight OpenAI wrapper.
• src/core/pipeline.ts – Functional pipeline framework for composing processing steps.
• src/core/logger.ts – Minimal structured logger, test-friendly.
• src/core/decorators.ts, env.ts, file-utils.ts – DX-enhancing utilities.
• src/core/ml-service.ts – Probabilistic multi-label classification base.
• src/types/env.ts, dataset.ts – Basic types to support module shape and validation.

Test Infra

• Added mocks: ollama-service.test.ts, pipeline.test.ts, logger.mock.ts.
• CLI preload support for testing IO (cli-progress.preload.ts, fs.preload.ts).

Project Setup

• Created tsconfig, .npmrc, and .gitattributes for TypeScript builds and Git LFS support.
• Defined base package.json, .gitignore, and coverage config.
• Introduced config stubs: categories.json, sites.json, ignore.json.

Config & Classifier Data

• Added early-stage config JSONs for future runtime logic.
• Added placeholder ai-zero-shot-classifier.d.ts for adapter development.