Emits the AI SDK UI message stream protocol from RubyLLM chats, so React front ends built with useChat talk to a Ruby backend: text, reasoning, tool calls and approvals.

Talks the native Anthropic Messages format to Bedrock InvokeModel / InvokeModelWithResponseStream, unlocking Anthropic-specific features that the Converse API cannot express — tool search with deferred tool loading, prompt caching via cache_control, and structured outputs — while keeping data ins...

An unofficial community extension for RubyLLM. It adds policy-bound, read-only browser Inspect and Extract tools with an offline FakeBackend. The Stagehand Ruby SDK adapter is present but unsupported for live operations until provider egress and cleanup are qualified. The first release also inclu...

Use Claude Code (and its subscription login) as a RubyLLM provider: streaming, images, PDFs, Office files, structured output and emulated tool calls over `claude -p`.

RubyLLM Code will provide an AI-powered coding assistant. This is a placeholder release - full implementation coming soon!

A RubyLLM tool that runs model-generated Ruby code inside a secure WebAssembly sandbox (SecurityBox), with explicit read-only and read-write folder mounts so agents can work on host project folders safely.

A RubyLLM provider that invokes the official Codex CLI, reuses local ChatGPT authentication, and maps RubyLLM structured-output schemas to Codex output schemas.

A comprehensive Ruby gem that provides Docker management capabilities through RubyLLM tools. Enables AI assistants to interact with Docker containers, images, networks, and volumes using natural language. Ported from DockerMCP to work directly with RubyLLM without requiring an external MCP server.

LLM evaluation engine for Rails.

Access Claude and GPT models through GitLab Duo via the RubyLLM interface. One provider, automatic routing to Anthropic or OpenAI backends.

A RubyLLM 2.x plugin. Every message is logged, a cheap model compresses the log into a binary tree of one-line summaries, and the chat sees a saved view of the tree that fits a budget the caller picks. The agent gets tools to zoom into any line, down to the original message. Based on Victor Taeli...

Validates and coerces unstructured LLM responses directly into rich, schema-validated Ruby objects with automatic self-correction loops.

Answers probability, choice, and score questions with any RubyLLM chat model, using parallel per-answer ratings or one-call typed distributions. Plugs into RubyLLM 2.1's Judge API and also runs on RubyLLM 1.13 through 1.16.

Declaratively define multiple modes for a RubyLLM chat and automatically route each user message to the appropriate mode. Each mode has its own instructions, tools, model, and reasoning effort, while all modes share the conversation history. Configure mode availability and fallbacks, and inspect ...

Drop-in Mongoid replacement for ruby_llm's ActiveRecord integration. Provides acts_as_chat, acts_as_message, acts_as_tool_call, and acts_as_model macros backed by MongoDB via Mongoid.

A Rails engine providing comprehensive observability for LLM-powered applications. Features include session tracking, trace analysis, prompt management, cost monitoring, and optional chat/agent testing UI (with RubyLLM integration).

Adds Pollinations AI support to RubyLLM — chat, image/video generation, TTS/music, and transcription via the Pollinations API. Installs as a standalone gem without modifying RubyLLM core.

Adds Apfel provider support to RubyLLM. Apfel is a macOS-only local server that exposes Apple's on-device Apple Intelligence model over an OpenAI-compatible API, letting RubyLLM talk to it like any other provider.

Adds a claude_code provider to RubyLLM. Prompts are written to the Claude Code CLI's stdin in batch mode and its JSON result is read back from stdout, so the agent (with its own tools, settings and resumable sessions) can be used wherever RubyLLM expects a model, e.g. `--model claude_code/opus` i...

Adds a codex provider to RubyLLM. Prompts are written to the Codex CLI's stdin in non-interactive mode and its JSON events are read back from stdout, so the agent (with its own tools, config and resumable threads) can be used wherever RubyLLM expects a model, e.g. `--model codex/default` in aia.