Create

Sign in to ReadmeX

or

How we handle your data: Privacy Policy

ReadmeX

Your AI reading companion

What would you like to catch up on?

Pick a briefing, or ask your own question.

Your community and people briefings stay personal to you.

Model Context Protocol

AI overview

Sign in and the AI will write an overview from our coverage.

Headlines · 152

  1. A Practical Guide to Building Production-Grade AI Agents

    Juejin’s first installment in an Agent development series explains how large language models differ from systems that can use tools, maintain state, and execute multi-step tasks. Using an in-app-purchase customer-support scenario, it connects Function Calling, Runtime loops, ReAct, Workflow, Planner-Executor, Multi-Agent patterns, Context Engineering, RAG, evaluation, observability, model gateways, security, approvals, and recovery into one engineering framework. The article also previews 12 follow-up lessons that build a runnable teaching project around order lookup and coin-credit troubleshooting.

    掘金 · 人工智能 · 🔥 8
  2. LangGraph full-stack project adds RAG and multimodal search

    The Juejin AI report describes the third installment of a LangGraph full-stack project, adding a lightweight local RAG knowledge base built with ChromaDB and an MCP project that generates webpages dynamically. It also consolidates online search, webpage generation, and knowledge-base retrieval in a search_node, requires human approval before web searches, and uses a multimodal model to recognize uploaded images and find similar images.

    掘金 · 人工智能 · 🔥 4
  3. Agentic AI Architecture: Integration, Security, and Scale Challenges

    InfoQ 中文 AI&大模型 examines how agentic AI systems can be incorporated into microservices and cloud-native enterprise architectures through embedded agents, MCP tools, A2A connections, and asynchronous messaging. The article highlights context engineering, domain boundaries, task-specific orchestration, fine-grained permissions, and observability as core architectural concerns while arguing that microservices and cloud-native systems will remain foundational.

    InfoQ 中文 AI&大模型 · 🔥 6
  4. Understanding the AI Agent Harness

    An article explains the AI Agent Harness as the execution and control layer supporting an agent, including model orchestration, context management, tool execution, state, security, and observability. Using a conceptual Codex workflow, it illustrates how a model proposes actions, a harness validates and executes them, and results are returned for further decisions. The article says the term has no fully standardized industry definition and outlines engineering concerns including context selection, permission boundaries, error recovery, and evaluation beyond response quality. Codex and Claude Code are mentioned as examples of AI coding tools, while the Codex workflow is explicitly conceptual rather than a complete account of its internal implementation.

    掘金 · 人工智能 · 🔥 3
  5. Practical LangGraph Guide to Integrating MCP and Managing Tool Selection

    A Juejin guide describes integrating Model Context Protocol (MCP) with LangGraph for Agent projects, recommending local stdio connections during development and environment variables such as MCP_URL to separate local and deployed resources. It recommends module-level lazy-loaded connections with cached tool listings, domain-specific graph routing, result truncation, graceful degradation, and progressive disclosure or retrieval to limit tool-selection errors as the available tool count grows. The author also warns about container-only DNS addresses, missing Node.js in some images, browser-dependent OAuth, and credentials leaking into images.

    掘金 · 人工智能 · 🔥 2
  6. MCP background task progress stalls after the original request closes

    A local test using Streamable HTTP and MCP SDK 1.30 found that notifications/progress reached the client for only one of five background task steps because the originating request connection had closed. The author showed that updating statusMessage and polling tasks/get retrieved all five steps, while callToolStream surfaced four of five updates. These results apply to the tested local configuration, and the author notes that behavior may differ under other transports and timing conditions.

    掘金 · 人工智能 · 🔥 1
  7. Pruning an Agent’s Tool Surface with Progressive Disclosure

    A report presents a runnable Python 3.12 implementation for reducing the tools exposed to an agent: filter candidates by task group, use a flat one-line index, load full schemas on demand, and downweight tools that repeatedly fail. Its tool_pruner.py uses a deterministic keyword selector and a character-based token estimate, so the comparisons demonstrate relative scale rather than model performance or actual billing. The report attributes Anthropic guidance that dynamic exposure can be worthwhile when tool definitions exceed 10K tokens, more than 10 tools are available, or tool-selection errors occur, while benefits may be limited with fewer tools or when every session needs all tools.

    掘金 · 人工智能 · 🔥 1
  8. Ruiyang urges enterprises to turn AI agents into governed business tools

    Ruiyang Information says enterprises should scale AI agents beyond isolated use cases and integrate them with existing ERP, knowledge-management and CRM systems. Its knowledge chief, Li Zhengquan, recommends the AI Technology Foundation’s four-stage maturity model, emphasizing data preparation, permissions, workflow integration, security and human oversight as prerequisites for business value. The article also cites industry research about agent-project failures and data-system barriers, while claims about multi-agent efficiency remain vendor assertions.

    科技新报 · 🔥 3
  9. Google details private MCP connectivity for Gemini Enterprise

    Google Developers Blog published a guide for connecting Gemini Enterprise agents to internal StreamableHTTP MCP servers hosted on Google Cloud Run. The reference architecture combines Egress Agent Gateway, Private Service Connect and FQDN-scoped Cloud DNS with dual service and user authentication, and includes a Python server, middleware and deployment scripts. Google says internal-only ingress blocks default public Gemini Enterprise calls with an HTTP 404 at Google’s edge.

    Google Developers Blog · 🔥 3
  10. BMW details how it governs hundreds of agents across thousands of developers

    AWS Machine Learning Blog describes how BMW Group’s R&D organization is building, governing, and operating hundreds of agents used by thousands of developers. The reported approach combines a central agent registry, framework-agnostic integration-layer controls, specialized agents, reusable infrastructure-as-code blueprints, and automated security and compliance checks. AWS and BMW say these practices speed deployment, reduce duplicate work, and maintain quality, although those benefits are not independently verified in the report.

    AWS Machine Learning Blog · 🔥 3
  11. Google launches Tool Retrieval preview for Gemini API

    Google launched Tool Retrieval in preview for the Gemini Interactions API, allowing agents to keep deferred tools out of the initial context and load their parameter schemas on demand. It supports Gemini 3.6 and later models, including Gemini 4 models, with server-side and client-side retrieval modes, and is enabled through a tool_search configuration with defer_loading set to true. Google says this can reduce input tokens, latency, and tool-selection errors, though the supplied reports provide no independent measurements of those benefits.

    Google AI · The Keyword · 🔥 2
  12. Watch the Claude agents session for Google Cloud

    Google AI on DEV says the on-demand “Build Agents with Claude Platform on Google Cloud” technical session is now available, with registration required. The session covers file handling, sandboxed code execution, custom skills, long-running task orchestration, tool use, production guardrails, and two deployment paths for Claude on Google Cloud. Google AI on DEV presents the Anthropic-managed path as using Google Cloud IAM, billing, and commitments, while the Google-managed path is described as supporting regional data residency, VPC Service Controls, and native integrations.

    Google AI on DEV · 🔥 0
  13. AWS publishes an AgentCore Gateway pattern for private API inspection

    AWS Machine Learning Blog published a tested architecture pattern for connecting Amazon Bedrock AgentCore Gateway to private APIs and MCP servers through a hub-and-spoke network with centralized firewall inspection. The pattern places an Amazon VPC Lattice resource gateway in the network account and includes six configuration decisions, validation checks, CloudFormation templates, test scripts, troubleshooting guidance, quotas, and an address plan. AWS built and tested it in us-east-1 using three member accounts and an AWS Organizations management account, and recommends adapting it to each organization’s account architecture.

    AWS Machine Learning Blog · 🔥 0
  14. A Practical Framework for AI Agent Tool Design

    The article organizes AI agent tools into five categories: perception, execution, collaboration, event triggering, and user communication. It compares dedicated code tools with Skills paired with a general executor, covering considerations such as parameter complexity, change frequency, model capability, and tool granularity. It also discusses MCP interoperability, layered organization, dynamic and progressive tool discovery, event-driven asynchronous agents, security risks from third-party MCP servers, and human-in-the-loop review. The author states that Manus, OpenClaw, and other general-purpose agents already provide most of the capabilities associated with Samantha in the film Her; this comparison is the author's assessment.

    掘金 · 人工智能 · 🔥 0
  15. Codex CLI 0.170.0 Adds Voice Calls and MCP Improvements

    Codex CLI 0.170.0 added WebRTC negotiation for real-time voice calls in its terminal interface and automatically refreshed authentication headers for OpenTelemetry HTTP exporters. The release also introduced experimental support for using same-named MCP servers across environments and fixed MCP OAuth credential races during cancellation, login, and logout, according to the release notes.

    OpenAI Codex CLI Releases · 🔥 0
  16. ScaleOps pitches autonomous orchestration for spiking AI agent workloads

    SiliconANGLE AI reports that AI agents can remain idle before bursting into compute-heavy activity, making static Kubernetes resource settings wasteful or inadequate. In a sponsored KubeCon interview, ScaleOps CEO Yodar Shafrir said the company adjusts allocations, replicas and node placement in real time, while allowing other workloads to use idle agent resources. ScaleOps claims these capabilities can cut cloud costs by 70% to 80% and make agent execution 90% more cost-efficient, but those figures are vendor claims.

    SiliconANGLE AI · 🔥 0
  17. AWS Agent Registry adds custom metadata schemas

    AWS Machine Learning Blog says AWS Agent Registry now supports organization-defined custom metadata schemas for agents, MCP servers, skills, and other records. Administrators can specify field names, types, allowed values, and required fields; the registry validates publisher-supplied metadata and supports review, discovery, and filtering.

    AWS Machine Learning Blog · 🔥 0
  18. AWS Quick sessions spotlight enterprise AI deployment at re:Invent 2026

    AWS published a guide to AWS Quick sessions at AWS re:Invent 2026, scheduled for November 30 through December 4. The agenda includes deployment, governance, security, analytics, memory and knowledge graphs, and enterprise scaling, with customer sessions featuring Allianz, Cox Automotive, Mondelez, DXC and Southwest Airlines. AWS also plans to stream more than 50 re:Invent sessions, including the CEO keynote and labs.

    AWS Machine Learning Blog · 🔥 0
  19. LoanPro deploys multi-agent loan-servicing assistant with AWS

    LoanPro built a production multi-agent loan-servicing assistant with AWS Partner AllCloud and the AWS Generative AI Innovation Center. The system uses Amazon Bedrock and Amazon Bedrock AgentCore to retrieve live borrower data, generate account summaries and recommend servicing actions. Its Data Retrieval Tool spans eight LoanPro API endpoints and synthesizes about 1,400 fields per account. AgentCore Gateway exposes the APIs as tools over the Model Context Protocol, handles authentication and manages connections, while AWS Lambda provides the integration layer. LoanPro said the deployment applies role-based access controls and compliance guardrails and records AI actions in its audit trail. AllCloud’s published case study reported customer call times reduced by up to 15%, but the supplied report provides no independent measurement methodology.

    AWS Machine Learning Blog · 🔥 0
  20. LinkBit offers a URL shortener for AI agents via MCP and API

    The available report describes LinkBit as a free URL shortener that AI agents can use through MCP and an API. It provides no additional technical details, usage or rate limits, pricing terms beyond the “free” description, or information about service reliability, security, or deployment status.

    Product Hunt · 🔥 0
  21. Multi-agent coordination patterns for manufacturing

    A tutorial organizes multi-agent coordination into two tiers: Tier 1 uses predefined flows, including sequential pipelines, conditional workflows and routers, while Tier 2 lets a model choose the path at runtime. It recommends beginning with deterministic Tier 1 workflows and reserving Tier 2 for tasks whose subtasks emerge during execution. The tutorial also identifies anti-patterns such as unbounded iteration, unstructured natural-language handoffs, premature adoption of multiple agents and unapproved real-world actions. Suggested safeguards include hard iteration and token limits, typed handoffs, observability, business rules and human approval before execution.

    AWS Machine Learning Blog · 🔥 0
  22. MoEngage Deploys Merlin AI on Amazon Bedrock AgentCore

    AWS Machine Learning Blog described how MoEngage built Merlin AI, a set of AI agents that operate on live platform data, using Amazon Bedrock, Amazon Bedrock AgentCore, and the Model Context Protocol. Its MCP Connector exposes existing OpenAPI-compatible campaign, analytics, segmentation, content, and flow capabilities as tools, with JWT validation, isolated sessions, and tenant-scoped access. MoEngage reports customer improvements including faster campaign launches and higher push click-through rates, but the figures are vendor-reported.

    AWS Machine Learning Blog · 🔥 0
  23. WrenAI becomes an open-source GenBI engine for AI agents

    The Canner/WrenAI repository describes WrenAI as an open-source generative business intelligence engine for AI agents, combining a semantic layer, an AI context layer, and governed text-to-SQL. Its Apache-2.0 engine includes an MCP server, CLI, and more than 20 connectors, while definitions are stored as reviewable YAML and Markdown in Git. The project also says Wren Engine merged into the main repository on May 7, 2026, while the earlier chat-first product was preserved as Wren GenBI Classic without new features or security fixes.

    GitHub Trending · Python · 🔥 0
  24. PrefectHQ Details FastMCP Python Framework

    PrefectHQ describes FastMCP as a Python framework for building Model Context Protocol servers, clients, and interactive applications. It automatically generates schemas, validation, and documentation from Python functions and manages transport negotiation, authentication, and protocol lifecycles for server connections. Its features include servers for exposing tools, resources, and prompts to LLMs, clients for connecting to local or remote MCP servers, and apps with interactive interfaces. Prefect says FastMCP 1.0 was incorporated into the official MCP Python SDK in 2024, that the standalone project receives one million downloads a day, and that some version powers 70% of MCP servers across all languages. The project also offers a TypeScript counterpart and FastMCP Gateway for centralized governance of MCP deployments, discovery, access, and auditing; these adoption and usage claims were not independently verified.

    GitHub Trending · Python · 🔥 0
  25. Why Hundreds of Tools Can Make an AI Agent Harder to Use

    A commentary argues that an AI Agent’s usability depends less on how many tools it can access than on how many tool descriptions it sees in each turn. It distinguishes capability formats—specialized functions, general-purpose command lines and Skill instruction files—from discovery strategies such as showing a compact catalog before expanding individual schemas. The article also cites a small Qwen3-4B comparison in which on-demand discovery reduced the amount of tool detail exposed and measured time, while both approaches completed all three tasks; the test does not show that the method improves accuracy or generalizes beyond that limited sample.

    掘金 · 人工智能 · 🔥 0
  26. AG-UI: An Open Protocol for Agent-User Interaction

    The AG-UI project describes AG-UI as an open, lightweight, event-based protocol for connecting AI agents with user-facing applications. Its stated features include streaming chat, bidirectional state synchronization, generative UI, real-time context enrichment, frontend tool integration and human-in-the-loop collaboration. The project lists integrations spanning frameworks such as LangChain, CrewAI, Microsoft Agent Framework, Google ADK, AWS Strands Agents and LlamaIndex; OpenAI Agents SDK, AWS Bedrock Agents, Nim and Flowise are marked as in progress. It also provides a reference HTTP implementation, a default connector and a command for creating an AG-UI application.

    GitHub Trending(每日) · 🔥 0
  27. Building a Read-Only MySQL MCP Server for WorkBuddy

    A guide explains Model Context Protocol (MCP) and demonstrates how to build a read-only MySQL server named mysql-explorer with FastMCP. The example exposes three tools, two resources and one prompt, then covers debugging with MCP Inspector, registering the server in WorkBuddy, testing that write queries are rejected, and deploying through stdio or Streamable HTTP. The guide includes implementation steps and manual verification instructions, but the supplied report does not provide independent test results across environments.

    掘金 · 人工智能 · 🔥 0
  28. Edera Adds Hypervisor-Level Security for AI Agents

    Edera announced an object-capability security model that separates AI agents’ task permissions from underlying credentials and enforces fine-grained controls at the hypervisor layer. Its Agent Runtime uses capsudo for scoped command execution and SPIFFE and SPIRE for workload identity; Edera says these initial capabilities are available. Credential-free API forwarding through capcurl and deeper orchestrator integration remain planned under Project Lunchbox.

    InfoQ 中文 AI&大模型 · 🔥 0
  29. Inside Pi’s Tool-Calling Implementation

    A source-code analysis examines Pi’s tool-calling flow, the second installment in the author’s Pi research series. It follows tool calls from the LLM’s structured request through execution in sequential or parallel mode, tool lookup, parameter adaptation and schema validation, beforeToolCall and afterToolCall hooks, cancellation checks, error handling, and the creation of tool-result messages for the next Agent Loop iteration. It also contrasts local tools with MCP tools, which the Harness accesses through an MCP client and server. The article does not identify the Pi version examined.

    掘金 · 人工智能 · 🔥 0
  30. NIST draft maps adversarial attacks on AI agents and MCP tools

    NIST released a public-comment draft classifying adversarial attacks against predictive machine learning, generative AI and agentic AI systems, according to iThome 台湾. The document covers tool poisoning, MCP server rug pulls, memory poisoning and instruction drift, and recommends cryptographic signatures, immutable version identifiers and sandboxed tool execution. The draft is open for public comment through December 18.

    iThome 台湾 · 🔥 0

Next →

Experience and discussion from the community

Share my Model Context Protocol experienceAsk about Model Context Protocol

cc-switch-cli: Manage Claude Code, Codex, Gemini, OpenCode, Hermes, OpenClaw, and Pi from one interactive TUI or scriptable CLI.

What problem it solves Developers using multiple AI command-line tools need a consistent way to manage providers, accounts, prompts, MCP servers, sessions, and configuration. Switching configurations manually across applications can also make it difficult to inspect usage, preserve settings, or use different providers in separate terminal sessions. Key features Provider management — Manage API con…

github.comhttps://github.com/SaladDay/cc-switch-cli

goose: A general-purpose AI agent that runs on your machine.

What problem it solves Users need an AI agent that can carry out tasks beyond producing code suggestions, including research, writing, automation, and data analysis. They also need access through desktop, terminal, and embeddable interfaces while retaining a broad choice of model providers and extensions. Key features Multiple ways to run — Use the native desktop application, a full terminal-orien…

github.comhttps://github.com/aaif-goose/goose

go-sdk: An implementation of the official Go software development kit (SDK) for the Model Context Protocol (MCP).

What problem it solves Go developers need packages for constructing MCP clients and servers while connecting them to transports and supporting protocol features such as OAuth. The SDK organizes these capabilities into importable packages and documents how supported MCP specification versions map to them. Key features Client and server APIs — The primary package provides APIs for constructing and u…

github.comhttps://github.com/modelcontextprotocol/go-sdk

Agentero: A unified research workspace for human and agent collaboration, combining PDF reading, structured context, literature management, notes, and writing.

What problem it solves Research context is often fragmented across PDF libraries, note applications, and separate model conversations, forcing researchers to move or copy information manually between tools. PDFs can also be difficult for models to interpret reliably, while common AI reading, translation, synchronization, and batch-import workflows require multiple plugins or repetitive manual acti…

github.comhttps://github.com/poco-ai/Agentero

txtai: An all-in-one AI framework for semantic search, LLM orchestration and language model workflows.

What problem it solves Developers need a unified way to build semantic search applications, ground language models with knowledge sources, and connect models into larger workflows. They also need support for different data types, models, deployment modes, and application interfaces without assembling each capability separately. Key features Unified embeddings database — The core component combines…

github.comhttps://github.com/neuml/txtai

voicebox: A local-first AI voice studio for cloning voices, generating speech, dictating into any app, and giving agents a voice you own.

What problem it solves Voice input and voice output are usually handled by separate cloud services, which means audio and transcripts leave the machine and the two halves of the voice loop do not connect. Voicebox aims to cover both directions locally: cloning voices from short samples, generating speech, dictating into any text field, and letting MCP-aware agents speak back in a cloned voice. Key…

github.comhttps://github.com/jamiepine/voicebox

agent-toolkit-for-aws: Help AI coding agents build, deploy, and manage applications on AWS.

What problem it solves AI coding agents need AWS-specific tools, knowledge, and guardrails to work effectively with AWS services. Teams using different coding agents also need a consistent way to configure AWS API access, guidance, and reusable skills. Key features AWS MCP Server — The managed MCP Server gives agents access to AWS APIs, isolated script execution, and current documentation through …

github.comhttps://github.com/aws/agent-toolkit-for-aws

jan: An open-source ChatGPT replacement for running LLMs locally with full control and privacy.

What problem it solves Jan provides a desktop application for downloading and running LLMs with local control and privacy. It also connects to supported cloud models and exposes an OpenAI-compatible local API for use by other applications. Key features Local AI Models — Download and run supported LLMs from HuggingFace on the local machine. The README lists Llama, Gemma, Qwen, and GPT-oss as exampl…

github.comhttps://github.com/janhq/jan

whatsapp-mcp: A Model Context Protocol server for interacting with a personal WhatsApp account through supported tools.

What problem it solves The server lets an MCP-compatible assistant search and read personal WhatsApp messages, contacts, chats, and message context. It also supports sending messages and media files to individuals or groups while keeping message history in a local SQLite database. Key features WhatsApp message and contact search — The MCP tools support searching contacts by name or phone number, l…

github.comhttps://github.com/lharries/whatsapp-mcp

figma-console-mcp: A design-system-focused Model Context Protocol server for Figma.

What problem it solves AI assistants need structured access to design-system data and controls in Figma to extract tokens, inspect components, create designs, and debug design-to-code differences. Teams also need deterministic checks for accessibility, design-code parity, and design-system health without imposing a specific framework or house style. Key features Design system extraction and token …

github.comhttps://github.com/southleft/figma-console-mcp

dbx: 100+ databases in 25 MB, available as a desktop app, Docker deployment, CLI, AI-assisted editor, and MCP Server.

What problem it solves Managing databases across many engines can require separate tools, while many database clients also depend on additional runtimes or bundled browser components. Teams and developers also need a consistent way to work with database connections across desktop, browser, self-hosted, CLI, and AI-agent environments. Key features Broad database coverage — Use one tool for more tha…

github.comhttps://github.com/t8y2/dbx