Whale – 98% Cache Hit with DeepSeek
Whale, an open-source AI coding agent built specifically for DeepSeek, achieves approximately 98% prompt cache hit rates, drastically reducing costs to pennies per session. The tool supports long cont…
Whale, an open-source AI coding agent built specifically for DeepSeek, achieves approximately 98% prompt cache hit rates, drastically reducing costs to pennies per session. The tool supports long cont…
A developer built a three-stage pipeline that auto-disables unused MCP plugins on a weekly schedule to prevent them from silently consuming context tokens in Claude Code. The system detects plugins en…
The 401 HTTP status code is not a rejection but a map that directs clients to the OAuth discovery documents, according to developer Prasham Trivedi's live production server mcplex. Trivedi's private M…
A developer discovered a production bug in their MCP server for Google's gemini-3.1-flash-lite-image model when a demo script revealed that the server's validation layer accepted thinking levels 'mini…
A user reports a ZeroGPU connection issue via the Anthropic MCP connector, with the error 'Anthropic proxy: MCP server connection lost' and 'Model not loaded'. An open GitHub issue (anthropics/claude-…
A developer's demo script for an MCP server that wraps Google's gemini-3.1-flash-lite-image model failed on its first run, revealing a production bug where the server's local validation accepted four …
An engineer tested Anthropic's Claude products—Claude Code, Claude Desktop, and Claude Cowork—and confirmed that all three use Client ID Metadata Document (CIMD) on user-initiated OAuth flows, contrar…
Developers benefit from AI agents that work inside shared systems like code repositories, while other roles get high-maintenance tools that sit outside the workflow, according to a blog post on the to…
XAgent is a task-first AI agent platform designed for teams, built around completing work rather than open-ended conversation. It features a server-side binary deployment, virtual file system for acce…
Agentproto 0.4.0 transforms the daemon from a launcher into a supervision surface, adding cron scheduling, blocking primitives, and a WorkflowRunner with explicit barriers. The release introduces sess…
Deja, a zero-dependency open-source binary, turns local agent conversation histories from Claude Code, Codex, and opencode into a searchable memory layer that works over SSH. The tool indexes gigabyte…
Hillnote, a local-first Markdown workspace with AI capabilities via CLI, Ollama, and MCP, has been launched on Show HN. The tool runs AI on-device even offline, supports documents, databases, kanban, …
A developer's guide demonstrates how to integrate an AI humanizer API into n8n, Zapier, and MCP-capable agents using the ToHuman API as a reference. The API rewrites AI-generated text to bypass detect…
A new open-source project called Aict reimplements 33 Unix coreutils with XML and JSON output, designed to eliminate the need for AI agents to parse human-readable plaintext. The tool, created by deve…
Five fail-open bugs were found in the policy engine of an MCP-mediated Slack agent, where missing data defaulted to permissive answers instead of safe ones. The bugs allowed a deny rule scoped to '/et…
A developer discovered that their MCP server, which wraps trusted APIs like GitHub and DEV.to, can inadvertently pass untrusted user-generated content into an AI agent's context, creating a data injec…
A developer introduces the Agent Capability Contract (ACC), a metadata layer that sits between tool connectivity and business authorization to govern AI agent operations. ACC adds machine-readable dec…
A developer released Inline Comments, a VS Code extension that lets users attach comments to selected text without modifying the file, storing them in a `.comments.json` file at the workspace root. Th…
Google has introduced the Agent2Agent (A2A) protocol, an open standard enabling AI agents built with different frameworks and vendors to communicate directly. The protocol defines JSON payloads, task …
A developer identifies six architectural patterns for Model Context Protocol (MCP) servers, ranging from direct API wrappers to composite tools and resource-based servers. The classification aims to h…