Tag
An unofficial Jev plugin for coding agents has been released, providing best practices, an API reference, and links to over 150 community projects, with evaluations demonstrating a 96% pass rate in coding tasks compared to other plugins.
Anthropic engineers outline a six-step framework for preparing for AI-driven code modernization, highlighting that as AI accelerates code changes, organizational processes become key bottlenecks.
The article discusses common errors in using AI agent traces and offers advice on proper implementation.
Analysis of 246 open-source repositories and 57 papers on agent harnesses highlights best practices like human-written context, limiting tools, and incremental testing to improve AI agent performance.
A tutorial by Brother Lan on best practices for using Meta's Muse AI Agent, emphasizing structured prompts, verification, and incremental permissions to avoid common pitfalls.
An AI agent reflects on four failures from running itself for months, emphasizing the need for independent monitoring, task verification, and caution against fabrication in persistent AI systems.
This blog post explains why developers must pass all unhandled messages to DefWindowProc in Windows programming, illustrated with a memory leak case involving RegisterDragDrop and RevokeDragDrop due to improper cleanup handling.
A discussion on best practices for managing API keys in AI agents, focusing on security measures like least-privilege access, key rotation, and preventing exposure of raw credentials.
The article argues that using GitHub's wiki for documentation is an anti-pattern and recommends using a docs folder with GitHub Pages for better version control and collaboration.
This article discusses the importance of evaluations in AI systems, explains why traditional testing is insufficient, introduces three main types of evaluations, and provides implementation suggestions.
The blog post proposes design guidelines for new programming languages, emphasizing fixes for common issues such as indented multiline strings, proper file path handling, consistent file extensions, and extensible syntax features.
This piece points out that many users mistakenly believe skills created with Codex or Claude function independently, but they often rely on local memory, as evidenced by failure on different machines.
The article discusses two camps in AI-assisted coding: using existing harnesses like Codex with tools and AGENTS.md, versus building custom orchestrator agents, with the author favoring the former as overhyped.
This article discusses how to manage code quality when using AI coding agents by adopting strategies like spec-driven development and test-driven development to reduce bugs and increase productivity.
GitHub shares evaluation practices for moving language model systems from prototype to production, addressing real-world challenges and metrics like precision and recall.
The author revised the success criterion for coding-agent evaluations by moving from requiring three perfect 88/88 runs to a criterion where each case must pass at least 2 of 3 runs, reducing randomness impact and improving stability.
The author advises logging detailed information in AI pipelines early on to prevent issues, sharing a personal experience where lack of documentation led to difficulties in debugging.
The article outlines 14 essential systems to build for ensuring the reliability and safety of AI agents, including identity management, access controls, and incident response protocols.
The article discusses Thomas Ptacek's rule for using LLMs as copyeditors, advising never to use LLM-suggested phrases to maintain originality and writing discipline.
The article provides style rules for coding agents to ensure clear, plain language output without jargon or bloat, covering vocabulary, sentence structure, and Markdown formatting.