Loop engineering — designing the cycle that runs a coding agent automatically — is the shift from hand-prompting every turn to building a system that finds work, runs the agent, checks results, and de…
Polygraph, a new Claude Code plugin and standalone CLI, uses LLMs to automatically generate formal models of stateful JavaScript code and exhaustively check every reachable state for rule violations, …
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 …
A developer built a single Python MCP server that exposes image-generation tools via the Model Context Protocol, allowing three different AI agents—a Google ADK agent, a Rust CLI client, and Claude Co…
A developer found that Cursor, an AI code editor, generated a login endpoint with a SQL injection vulnerability (CWE-89) by interpolating user input directly into a query string. The developer warns t…
A developer audited an AI-assisted refactor of the open-source CLI tool YouMindAG and found 46 undefined references across 7 files, silently breaking 7 of 14 commands. The refactor reduced a 1,920-lin…
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…
A developer used OpenAI Codex to add a 'brainless' pricing block to a landing page, running the command 'bunx shadcn add brainless/pricing' and updating app/page.tsx with the <Pricing tiers={TIERS} />…
A developer with 17 months of intense Claude Code experience shares advanced tips including auto mode, remote control from a phone, and a three-step workflow of investigate, design, and implement. The…
Software is becoming disposable as AI tools like Claude Code make new code easier to write than old code to maintain, according to Seth Godin. The network, however, remains the enduring asset, making …
A developer's analysis of the AI-agent ecosystem reveals that model launches are the strongest acquisition event for agent tools, with openai/codex doubling claude-code's star pace within hours of GPT…
A developer released ai-credit-scrub v1.1.0, a free, offline-first Go CLI that removes explicit AI-agent credit lines from commit messages and pull request bodies before they reach Git history or GitH…
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…
Sierra has built Pinecone, an internal cloud-based agent that captures and scales employee workflows across the company, turning individual improvements into organizational knowledge. The system provi…
LLM-Wiki v0.16.0 introduces Query Lite, a compact read-only query protocol that reduces Claude Code wiki query instructions by 72.54%, alongside token benchmarks and reproducible benchmark lanes for C…
A developer at Agent Island describes why a coding-agent completion event is insufficient for accurate state tracking. The team built a turn-based status model with deduplication keys and semantic tim…
AI control research must expand from models to agent harnesses as frontier labs adopt harnesses with skills, memory, subagents, and external services by 2026, according to a LessWrong analysis. Claude…
A developer compared Playwright MCP and Playwright CLI for use with Claude Code, finding that CLI uses approximately 26,000–27,000 tokens per multi-step UI test versus MCP's 114,000 tokens, a differen…
Fin has published a curated set of Claude Code skills and hooks used by hundreds of its engineers, alongside a framework distinguishing effective from ineffective skills based on its internal experien…
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…