Static analysis for LLM call sites. Find what your prompts cost before you ship them.

A high-performance LLM proxy router for load balancing, monitoring, and managing API requests to various language model providers

Plug-and-play LLM gateway + consolidated usage & cost dashboard for every AI project you run - any language, one dashboard, one command.

Simple RAG Chat using Upstash

OpenTelemetry SpanProcessor for atrib's verifiable action layer. Emits signed records from OpenInference-shaped spans.

PeakInfer MCP Server for Claude Desktop and Claude Code

Read your agent's run back. A harness-neutral CLI (Claude Code + OpenAI) that turns an agent transcript into a flight review + a skill to extract — so the next run is better, not just observed.

Local proxy that intercepts Anthropic API calls, monitors token spend, and optionally routes to cheaper models when budget runs low.

Open-source AI model gateway core for one-key, multi-provider routing across Hasna apps and self-hosted deployments

PeakInfer MCP Server for Claude Desktop and Claude Code

One consistent function to call any LLM provider — Anthropic, OpenAI, Gemini, Moonshot, and more.

CLI tool to auto-generate OpenCode configuration by fetching models from providers

Real-time cost awareness for MCP agent workflows

MCP server that advises LLM model routing & cost BEFORE the call — estimate_cost, recommend_model, should_offload_local. Pure logic, no API keys, $0 to run.

Voight observability for the Vercel AI SDK. OpenTelemetry SpanExporter that ingests `experimental_telemetry` spans from `streamText`/`generateText`/etc. into your Voight dashboard — prompts, tokens, tool calls, cache, latency, errors, all providers normal

Local dashboard for LLM API rate limits across Claude, GPT, Gemini, OpenRouter, Groq, Fireworks, Mistral and Together. Zero dependencies.

Token-aware rate-limit queue proxy for the Anthropic Claude API. Agents never see a 429.

Drop-in observability and cost tracking for LLM API calls. Wrap your provider client, log every call with cost/tokens/latency to JSONL, query via CLI or MCP.

A fuse box for your AI bill. Drop-in spend protection and runaway-loop circuit breaker for Claude and OpenAI agents. Includes Python and Node SDKs plus a CLI.

Tunerkit Node SDK: E2E development of AI Agents