Fast-Editor/Lynkr

Streamline your workflow with Lynkr, a CLI tool that acts as an HTTP proxy for efficient code interactions using Claude Code CLI.

What it solves

Lynkr is an LLM gateway designed to reduce token costs and latency for AI coding tools. It prevents users from being locked into a single provider and stops tools from sending raw, inefficient token streams, which often leads to high bills and wasted subscription capacity.

How it works

Lynkr acts as a proxy between AI tools (like Cursor, Claude Code, or Codex) and LLM providers. It optimizes traffic through several mechanisms:

  • Tier Routing: Automatically routes requests to different models based on complexity (Simple, Medium, Complex, Reasoning). Simple tasks go to cheap or local models, while complex ones escalate to powerful cloud models.
  • Token Compression: Uses TOON/RTK compression to reduce tokens on large JSON tool results (e.g., grep or file reads) by up to 84%.
  • Semantic Caching: Stores repeated queries to serve them in under 300ms with zero billed tokens.
  • Wrap Mode: Allows AI coding assistants to use these routing and compression features without requiring code changes to the assistant itself.
  • Learning Loop: Uses a kNN router and bandit system to calibrate routing thresholds based on actual traffic outcomes.

Who it’s for

Developers using AI coding assistants (Cursor, Claude Code, Codex, Cline, Continue) who want to lower their API costs, utilize local models via Ollama or llama.cpp, and optimize their subscription usage.

Highlights

  • Multi-Provider Support: Compatible with 14+ providers including Ollama, OpenAI, Azure, AWS Bedrock, and DeepSeek.
  • Zero Code Changes: Integrates via base URL overrides in existing tools.
  • Real-time Dashboard: Built-in telemetry for tracking spend, savings, routing accuracy, and provider health.
  • Streaming Support: Native passthrough and cross-format transformation for SSE streaming.
  • Optional Code Intelligence: Integrates with Graphify for AST-based complexity scoring to improve routing decisions.

Related

  • Project
    wevm/incur

    incur is a TypeScript library that lets you build command‑line tools that are easy for LLM agents to discover and use. It provides typed command definitions (via Zod), token‑efficient TOON output, built‑in MCP and skill‑file generation, and can wrap or expose HTTP APIs as CLI commands.

  • Project
    agentjido/req_llm

    ReqLLM is an Elixir library that unifies calls to dozens of LLM providers (OpenAI, Anthropic, Azure, Google Gemini, etc.) behind a single, high‑level API. It offers functions like `generate_text/3`, `stream_text/3`, and `generate_object/4` that work across 1,200+ models, handles streaming via Finch, validates structured outputs, tracks token usage and estimated cost, and provides built‑in Telemetry. Keys are managed automatically from env files or in‑memory storage, and the library includes extensive provider guides and a Discord community for support.

  • Project
    theopenco/llmgateway

    LLM Gateway is an open‑source API gateway that provides a single OpenAI‑compatible endpoint for routing requests to multiple LLM providers, centralising API‑key management and offering built‑in usage and performance analytics. It can be used as a hosted SaaS service or self‑hosted via Docker.

  • Dispatch
    Lowfat: Reducing LLM Token Costs via Pluggable CLI Filtering

    Lowfat is a lightweight, local-first CLI tool designed to strip noise from command output to reduce token consumption for AI agents.

  • Project
    diegosouzapw/OmniRoute

    OmniRoute is an open‑source AI gateway that aggregates 350+ LLM providers (including 150+ free tiers) behind a single OpenAI‑compatible endpoint. It auto‑routes requests, applies token‑compression, tracks free‑tier quotas, and works out‑of‑the‑box with zero configuration.