v0.1.1

⚡ What's New in v0.1.1

This release introduces native integration for the Vercel AI Gateway provider adapter (@ai-sdk/gateway), expands package metadata for ecosystem discoverability, and delivers a comprehensive overhaul of the README documentation.

---

🚀 Highlights & New Features

🌟 Vercel AI Gateway Provider Adapter (vercelgateway)

• Unified Gateway Routing: Added support for the vercelgateway/ provider prefix, allowing requests to be routed through Vercel AI Gateway using @ai-sdk/gateway.
• Preserved Model ID Routing: Supports nested model identifiers (e.g., vercelgateway/openai/gpt-4o or vercelgateway/anthropic/claude-3-7-sonnet) with automatic default base URL fallback (https://ai-gateway.vercel.sh/v4/ai).

📚 Comprehensive Documentation & Ecosystem Discoverability

• Expanded README Guide: Completely overhauled the documentation with clear architecture rules (First-Slash Parsing, mandatory credentials, multi-tenant concurrency), explicit unsupported features (image generation, environment variable singletons), and AI agent guidance.
• Enhanced Package Metadata: Added comprehensive npm keywords, badges, homepage links, repository topics, and issue tracker URLs to package.json.

---

🛠️ Bug Fixes & Improvements

• Registry Expansion: Updated the internal provider registry (src/providerRegistry.ts) and test suites (tests/providerResolution.test.ts) to validate all 13 built-in provider adapters.
• Dependency Update: Added @ai-sdk/gateway (v4.0.121) to dependencies.

---

📝 Commits

View full commit log

• d4b3438 feat: add Vercel AI Gateway provider adapter and bump version to 0.1.1
• a9a2138 chore: add comprehensive SEO metadata, keywords, badges, and repo topics
• 1e6fa32 docs: enrich README with explicit unsupported features and agent guidance

v0.1.0

🚀 @amaranthic/greg-sdk v0.1.0

Executive Summary

We are thrilled to announce the initial release of @amaranthic/greg-sdk (v0.1.0), a powerful, drop-in runtime built on top of AI SDK v7. This release introduces comprehensive multi-provider abstraction, robust type safety, canonical runtime errors, and a full suite of AI generation capabilities.

🚀 Highlights & New Features

• Comprehensive Multi-Provider Adapter Architecture: Integrated adapters for top AI providers, including OpenAI (via Responses API), Anthropic, Google, DeepSeek, xAI, Groq, Mistral, Cohere, Together AI, Alibaba Cloud, OpenRouter, and custom OpenAI-compatible endpoints.
• Unified AI Runtime Surface: Implemented core runtime execution functions—generateText, streamText, generateObject, and streamObject—re-exporting an AI SDK v7 compatible surface.
• Advanced Model Resolution & Credential Validation: Added modelResolver supporting first-slash parsing (provider/model) and mandatory credential validation.
• Canonical Runtime Error Classes: Established strongly typed, structured error classes for runtime exceptions and error handling across provider integrations.
• Robust Testing & CI/CD: Shipped with a full suite of 6 high-coverage test files utilizing the Bun test runner alongside an automated GitHub Actions release workflow configured for npm OIDC trusted publishing and provenance.

🛠️ Bug Fixes & Improvements

• TypeScript & Typing: Added strict tsconfig.json settings, derived TypeScript types, interfaces, and strict JSDoc requirements.
• Dependency Locking: Locked exact versions for top AI provider packages and AI SDK dependencies.
• CI Environment Hardening: Configured OIDC trusted publishing environment settings and clean .npmrc handling for secure npm releases.

📝 Commits

View full commit log

• fc6e2e0 ci: set environment to TOKEN to match npm trusted publishing configuration
• e95191e ci: ensure npm@latest and clear .npmrc for OIDC trusted publishing
• 0e12558 ci: use pure npm OIDC trusted publishing without registry-url and tokens
• 24b4952 ci: add automated release workflow with ant AI release notes and npm provenance publishing
• 1d4c1d9 docs: update tagline, documentation, MIT license, and interactive chat example
• 071b1a4 test: add full suite of 6 high-coverage test files for Bun test runner
• 72d2a8d feat: complete public index entrypoint re-exporting AI SDK surface and Greg runtime
• f32198d feat: add runtime functions generateText, streamText, generateObject, streamObject
• f71d92a feat: add modelResolver with first-slash parsing and mandatory credential validation
• ea030a5 feat: add central providerRegistry in src/
• c7f4a2a feat: add Alibaba Cloud provider and OpenRouter provider adapters
• ae8c323 feat: add Together AI provider adapter
• a0e6427 feat: add Cohere provider adapter
• a3b7cc1 feat: add Mistral provider adapter
• 1589a6b feat: add Groq provider adapter
• 60a7d95 feat: add DeepSeek provider adapter
• 5de6f5e feat: add xAI provider adapter
• 96ff33b feat: add Google provider adapter
• f52c7c9 feat: add Anthropic provider adapter
• 575e3d6 feat: add OpenAI-compatible provider adapter
• 4892173 feat: add provider base adapter and OpenAI adapter using Responses API
• 01e4d77 feat: add derived TypeScript types and interfaces
• b5b4c8f feat: add canonical runtime error classes
• f87c727 chore: add top 10 provider packages with exact versions locked
• 9e54ee6 chore: install and lock ai, typescript, and @types/node
• 854a232 chore: add full strict tsconfig.json
• 5689685 chore: initialize package.json without dependencies
• 28a98cd docs: add strict casing standards and JSDoc requirements to AGENTS.md
• c9ab266 docs: add README defining Greg SDK as a runtime on top of AI SDK v7
• fc61640 chore: add .gitignore and stop tracking PLAN.md
• 87933a1 chore: initial commit with plan, guardrails, and agent skills