LLM integration layer for Formloom - prompt fragments, tool definitions, parser, and return-path helper for OpenAI, Anthropic, Gemini, Mistral, and Ollama

Connect Sisu to OpenAI and OpenAI-compatible APIs with tools and streaming support.

Port account-level LLM configuration — custom instructions, Projects/Custom GPTs, saved memory, and request-level settings — between Claude, ChatGPT, and Gemini, and get a report of exactly what did not survive the trip.

wllama provider for @localmode - GGUF model inference via llama.cpp WASM with metadata inspection and browser compatibility checking

Expose ChatGPT web sessions as a local OpenAI-compatible Responses API.

SDK for integrating with AI Providers plugin

Dynamic tool loading for opencode — Kimi K3 tool-search pattern, adapted to opencode's plugin runtime. Collapses noisy tool descriptions into a searchable catalog so the LLM only sees relevant tools per turn.

Mistral AI provider integration for Quilltap

Harness de evals de seleção de tool para servidores MCP: extrator de catálogo (fake McpServer), validação de fixtures, scorer offline (top-1/top-k/por-área), gate de acurácia e runner via Anthropic Messages API.

Structured JSON output from 12 LLM providers with a single function call. Type-safe, retry-resilient, zero boilerplate.

Proxy service that translates Anthropic API format to OpenCode Go API format. Use OpenCode models with Claude Code and other Anthropic-compatible tools.

A TypeScript SDK for interacting with Ollama Proxy, supporting multiple LLM providers with fallback capabilities

OpenAI function-calling and Anthropic tool-use schemas for the CRYPTYX x402 API. Drop-in tool definitions for any GPT- or Claude-based agent that needs institutional-grade crypto conviction, walk-forward-validated signals, and evidence-graded trigger eval

Chat Model node for n8n — use a private LLM gateway as the LLM behind an AI Agent or LLM Chain (with tool calling).

Drop-in OpenAI-compatible tool calling for LLMs that don't support function calling natively. Injects tools via prompt, parses tool calls from text, and runs a full agentic loop — with streaming support.

Proxy for NVIDIA NIM API that fixes GLM-5.1 tool_call streaming bugs (numeric IDs, missing function.name, ID instability)

OpenClaw Dify Auth provider plugin

CLI bridge between Eqho AI platform and promptfoo evaluations

Local proxy: Anthropic Messages API → Codex Responses API using local Codex OAuth

Runtime governance and execution evidence for AI agents.