OpenAI SDK adapter for llm-ports. Implements LLMPort and EmbeddingsPort. baseURL support covers 10+ OpenAI-compatible providers (Azure, Groq, Together, Fireworks, DeepInfra, Perplexity, Cerebras, LiteLLM proxy).

Provider-agnostic LLM port interface, registry, cost gating, content blocks, validation strategies. The foundation of llm-ports.

Direct @anthropic-ai/sdk adapter for llm-ports. Implements LLMPort with prompt caching, vision, and tool use support for Claude models.

Reusable cognitive operation factories for llm-ports: classify, score, draft, summarize, extract, plan, analyze. Configure once at definition time, call many times.

Subprocess-driven Aider CLI adapter for @llm-ports (Shape A passthrough governance).

Ollama native adapter for llm-ports. Local LLMs via the Ollama daemon. Implements LLMPort, EmbeddingsPort, and ModelManagement (list/pull/delete/health).

Google Gemini adapter for llm-ports — native @google/genai SDK integration with bundled pricing, content-block translation, validation repair, and image-size + URL validation.

Vercel AI SDK adapter for llm-ports. Migration path for users already on @ai-sdk/*. Bring your own Vercel-configured models; the adapter handles the LLMPort plumbing.

Standalone data contract for LLM observability: event envelope, correlation model, lifecycle events, error info, cache stats, prompt fingerprint, evaluation refs, capture policy, W3C Trace Context + Baggage. Zero runtime dependencies on @llm-ports/core so

OpenTelemetry semantic-conventions bridge for @llm-ports/observability-contract. `createOtelSink({ tracer, meter? })` maps every lifecycle event into gen_ai.* spans + metrics per the OTel gen_ai semconv. Zero runtime dependencies; consumers pass their own

Subprocess-driven adapter for OpenAI Codex CLI. Runs `codex exec --json` as a subprocess, parses its JSON event stream, and emits @llm-ports contract events. Shape A: passthrough governance (operator supplies OpenAI credentials to codex directly; @llm-por

Durable storage for post-hoc evaluations (LLM-judge scores, human annotations, rule-based verdicts) keyed on the @llm-ports/observability-contract EvaluationRef shape. Ships an in-memory store (default) and an opt-in SQLite writer (peer-dep on better-sqli

Route a LiveKit Agents voice agent through an @llm-ports LLMPort. LiveKit already abstracts providers; this adds what its LLM layer lacks: fallback chains, USD cost ceilings, per-provider budgets, and task routing. One constructor swap; the agent's tools,

Codemods for migrating consumer code across @llm-ports/* alpha releases. Conservative, best-effort, runs offline.

wait-on is a cross platform command line utility and Node.js API which will wait for files, ports, sockets, and http(s) resources to become available

A simple tool to find an open port on the current machine

A high-performance JavaScript 2D/3D polyline simplification library

Normalize browsers' default style

Boolean operations on polygons (union, intersection, difference, xor)

LZ-based compression algorithm