llm

A terminal-based agentic coding harness for engineers, optimized for z.ai GLM models and built to hot-swap across multiple LLM providers mid-session.

Open-source agent harness that starts from a tiny loop, gives the model one exec tool, and stores the agent’s state and sessions inside its own directory. It includes setup commands, model configuration notes, and design rationale.

A small Rust-based LLM proxy for forwarding chat, messages, and responses requests to providers with gateway-key handling and optional protocol translation between Anthropic and OpenAI styles.

Open-source Python library that lets developers integrate an LLM into the Python runtime so missing attributes, methods, and typed values can be inferred on demand with confidence scores.

A Rust-based coding agent built for speed, low memory use, and Claude Code compatibility. The project highlights support for skills, hooks, MCP, plugins, sub-agents, and multiple LLM providers, with terminal, headless, and IDE-facing entry points.

A curated GitHub resource mapping AI token costs, waste, and efficiency practices for agentic coding. It combines an awesome list with editorial notes, practices, concepts, claims, and copy-paste setups.

NakshGuard is a self-hosted reverse proxy for OpenAI and Anthropic chat APIs that detects and blocks runaway AI agent loops before they waste tokens or API spend. It includes per-agent session tracking, configurable thresholds, shadow mode, and hot config reload.

Local-first governance layer for AI systems that monitors prompts and responses to block secrets, PII, and runaway LLM spending before requests leave the machine.

Python library for managing LLM conversation history with cache-friendly message ordering, dynamic context, and compaction.

A macOS desktop shell where apps are generated by Claude into interactive HTML mini-apps, with local system features like windows, dock, file storage, notifications, and persona-based themes.

Lathe generates multi-part, hands-on technical tutorials from prompts and provides a local UI plus CLI workflow for working through them with LLM-assisted skills.

A Python filter layer for LLM apps that evaluates responses against a Lean-verified Scherf logic backend and can pass, flag, correct, or block output.

A PyTorch reference implementation of DeepSeek Sparse Attention from the LLMs-from-scratch project, including a lightweight indexer, top-K token selection, and a GPT-style model with KV-cache support.

MagesticAI is an open-source, browser-based platform for managing AI coding tasks with multi-agent orchestration, task boards, terminal access, and code review. The README shows installation steps, Docker deployment, supported platforms, and a concrete local setup path.

A walkthrough for building and deploying an AI agent on Truuze with the AaaS dashboard, including setup, testing, connector configuration, and the platform’s escrow-based transaction flow.

UltraCompress is a Python-based compression tool for large language models. The repo describes lossless 5-bit transformer compression, verification via SHA-256, a CLI on PyPI, and published model packs on Hugging Face.

OpenFlow is an open-source visual builder for multi-agent AI workflows, built on the opencode engine. The page shows the project’s purpose, setup steps, runtime details, and how the engine and canvas are launched together.

Open-source CLI that analyzes AUDR traces to find avoidable LLM calls, wasted tool usage, context growth, and other execution inefficiencies in agent runs.

CloudGrip is a self-hosted AI proxy gateway and dashboard for controlling LLM API spend, enforcing budget caps, and reviewing request telemetry.

An interactive browser demo that visualizes attention in LLM text generation, letting users hover or tap tokens to inspect which earlier tokens influenced them.

Scalattice is an inference API and hosting platform that lets developers point the OpenAI SDK at its OpenAI-compatible endpoint with minimal code changes. The page includes Python and curl examples, model IDs, pricing references, and basic troubleshooting.

An open-sourced pipeline that uses Gemini, GoPro footage, Garmin .fit telemetry, Strava segments, and manual labels to automatically build cycling highlight reels.

Reference implementation for using Linux Pressure Stall Information to trim an LLM KV cache under memory pressure. The repo includes requirements, basic usage commands, a simulator, a llama.cpp runner, and benchmark scripts with example results.

A browser-game playing agent that uses an LLM to train per-game rules and then makes live decisions through either a hosted engine (Jev) or a local engine (Laya).