context-caching

Improving Spring Boot test efficiency

Speed Up Your Spring Boot Test Suite ⏱️

MCP server that uses Gemini's context caching as extended memory for AI assistants. Load GitHub repos, URLs, PDFs.

ArchonCLI is an autonomous AI code architect powered by Gemini 3. It uses Tree-Sitter for syntax-aware indexing and RAG to provide deep insights into complex codebases via a modern TUI or scriptable CLI.

RACS (Remote Agent Context Store): prefix-cache management for production agents. Stability-aware prompt planning, provider-faithful cache directives for Anthropic, OpenAI, Gemini, Bedrock and more, TTL keep-warm scheduling, prefix-drift detection, and hit-ratio and savings analytics. Zero dependencies, TypeScript, edge-ready.

Eine smarte KI-Assistentin für ioBroker (inspiriert durch Eureka). Verbindet Telegram, DeepSeek-KI, MQTT (Tasmota) und Websuche mit Cache-Optimierung.

Global Pi extension: context caching optimizer, .claudecodeignore automation, repo-hop detection, compact watchdog, and warm KV-cache routing

Multimodal orchestration for LLM APIs. Source patterns, context caching, and structured output for text, PDFs, images, video, and YouTube - so you don't manage the complexity yourself.

🤖 Structured Agentic Instructions and Code Templates optimized for Gemini 2.5.

AI coaching system that scores running biomechanics in real-time using Gemini 2.5 Flash Lite with context-cached research papers — 3 Cloud Run microservices

2M-Token Monorepo Intelligence, Context Caching & Blast Radius Analyzer

Native Google GenAI SDK integration for LangGraph. Context caching (5x cost reduction), bidirectional message translation, tool auto-conversion. Drop-in replacement for ChatGoogleGenerativeAI in LangGraph nodes.

Unified orchestration workbench for AI coding agents (DeepSeek Harness / Claude Code / Codex) with cross-session context caching to cut token costs

Local-first agent workspace: Electron desktop IDE and Python/FastAPI connector, DeepSeek agent with subagent workers, Codex and Claude adapters, Telegram, AutoRig image/video/3D generation, self-updating supervisor.

Semantic caching layer for multi-agent LLM systems. Stop paying your LLM to recompute the same thing twice. MCP-native.

🧠 Pebble: A local-first, append-only memory substrate for AI agents. Featuring an MCP server, context hot-caching, and Obsidian-compatible vault rendering.

基于 QuantumNous/new-api 的 Gemini 显式缓存扩展：缓存生命周期管理、固定渠道与密钥绑定、独立计费。AGPL-3.0。

A lightweight, engineering-first LLM framework for FastAPI featuring session-based context caching and reliable output validation.

Cache-Augmented Generation (CAG) demo with the Gemini API — load documents into a context cache once, query at a ~90% token discount. Live savings dashboard + CAG-vs-full-context compare mode.

Agent harness em Python puro com context caching medido — 79,3% de economia (benchmark real com Gemini, 5 repetições por condição)