"MCP server for code understanding and agent context initialization" matching MCP servers:

A minimal MCP server that exposes Gemini's web and multimodal understanding (video, audio, document, image) through five dedicated tools.

MCP server for multimodal understanding and object grounding (bounding boxes) across images, videos, and documents, with support for multiple AI providers (Zhipu GLM-V, OpenAI GPT-4o, Anthropic Claude, or any OpenAI-compatible endpoint).

Provides persistent codebase context caching via Gemini API, enabling fast, cost-effective repeat queries for MCP hosts like Claude Code.

Integrates Xiaomi MiMo's multimodal API to enable understanding of images, audio, and video through natural language prompts.

An MCP server that uses Google's Gemini API to analyze videos and convert them to text descriptions that Claude Code can understand and act upon.

MCP server for deterministic context preflight, token counting, and credential redaction for coding agents, with a free local tier up to 12,000 characters and paid HTTP fallback for larger inputs.

A multi-persona code review tool that leverages AI agents with distinct expertise (Codex, Toxic Architect, Accessibility Expert) to analyze code quality through the Model Context Protocol.

Enables AI agents to search, filter, and monitor curated AI model API free credits and low-cost token promotions through eight tools, four resources, and three prompts — including finding deals expiring soon, checking eligibility and blocked conditions, and producing daily briefs.

Pay-per-task AI agent for writing, research, code, DeFi & blockchain. Pay in USDC on Base or Solana. Supports A2A, MCP, x402 and Agentmail protocols.

Enables AI agents and MCP-compliant clients to dynamically shrink long-horizon context windows, pruning verbose tool outputs, HTML boilerplate, and semantic redundancy by roughly 40-70% without frontier LLM round-trips. Runs deterministically as a zero-dependency Python stdio server, accepting a payload plus optional compression and risk-bound options.

MCP server that cuts cloud LLM costs 36-42% by indexing context locally and giving agents precision retrieval tools instead of raw context dumps.

Translates natural language commands into structured geometric operations for a simulated CAD engine using LLMs. It enables users to create shapes, modify dimensions, and perform 3D operations like extrusion through the Model Context Protocol.

An MCP server that lets Claude Code delegate coding tasks such as code generation, review, refactoring, and test writing to a local Ollama instance running a Qwen3 model.

An MCP server that recommends UI components and utility methods from private code repositories using AI to enhance code reuse and development efficiency.

Connects Kimi Code with Claude Code, enabling Claude to delegate bulk codebase reading to Kimi (256K context) for cost savings, while Claude focuses on reasoning and code edits.

MCP server for agent reputation staking — trust through skin-in-the-game. Agents deposit stakes as trust collateral, with slashing for bad behavior and reputation building over time.

Enables planning with Claude Code and high-speed code generation using Cerebras, avoiding API limits while producing quality code with visual diffs.

Long AI conversations fail in predictable ways. Context-First fixes all four: Failure Mode What Goes Wrong Context-First Solution Context Drift AI forgets earlier decisions and intent as the conversation grows context_loop + detect_drift continuously re-anchor every turn Silent Contradiction New inputs silently overrule established facts — the AI doesn't notice detect_conflicts compares every inp

A dual-mode MCP server and CLI agent for AI-powered tasks including weather, news, file management, web scraping, and data processing via LangChain.

Imports private ChatGPT conversations into MCP clients like Codex and Claude Code. Provides tools to fetch remote conversation content and serve cached context via local stdio.