context-window-optimization

A super light-weight embedded code search engine CLI (AST based) that just works - improves speed and efficiency for coding agent 🌟 Star if you like it!

14-stage Fusion Pipeline for LLM token compression — reversible compression, AST-aware code analysis, intelligent content routing. Zero LLM inference cost. MIT licensed.

Verified code context for agents

Give your Claude Code Agent Teams a memory. Auto-injects role-specific context into every new teammate — your team never starts blind again.

ISONGraph is a multi-language property graph library designed specifically for LLM and AI applications. It provides the most token-efficient way to represent graph data in LLM context windows while maintaining high accuracy for graph-based reasoning tasks.

Baton coordinates coding agents. Each worker handles a focused task, and one orchestrator checks the result.

Light-weight, query-aware compression for high-signal LLM prompts.

A system architecture proposal for infinite AI conversation continuity, solving the context window limit through a five-component memory framework: sliding hot window, conversation graph, state document, memory manager, and pre-fetch engine

Field projection layer for MCP tools — let AI agents request only the fields they need, reducing context window usage by up to 95%

ESLint for MCP servers — a read-only CLI that audits MCP tool definitions for token cost, description quality, and schema hygiene before your agent burns its context window.

A structure-aware, local prompt compressor and context optimizer for LLM pipelines.

A chatbot with RBAC and Human-in-the-Loop used for booking purposes covering the research, plan formation, booking, payment. This is built for multiple users like - Booking customers, internal teams (finance, customer support etc..). Built with multi-agent two-layer supervised architecture in LangGraph powered by LLMs including openAI,Gemini, Ollam

Automate content research, card news, images, voice, and video from one prompt with an end-to-end Claude Code content pipeline

Exact dynamic output budgeting for Hermes Agent

An expense tracking app using Agentic Coding

Local LLM inference benchmarker. Measures TPS, TTFT, and VRAM pressure across context sizes — from 2K to 256K tokens. Works with llama.cpp, Ollama, LM Studio, and any OpenAI-compatible server.

Prepares projects for AI by generating compact manifests. Send 10-50× less data to LLMs. Python CLI, no dependencies.

Cuando una sesión larga de Claude Code se resume sola, ¿qué se ha perdido por el camino después de compactar? Esto lo cuenta leyendo el JSONL de sesión: ¿Cuántos de los ficheros que has usado sobreviven al resumen?

Graph-based GitHub repo analyzer that cuts LLM context costs 50–70% — builds a dependency graph via AST parsing, then selects only the most relevant files for a query instead of dumping the whole repo. FastAPI backend + force-directed graph visualizer.

Non-destructive skill manager — compiles the right skill loadout per project, then demotes and hides unused skills to keep your context window lean. Never deletes without your approval.