Full-text search across 60806 articles. Combine with topic and date filters; results sorted by relevance.
Biscuit, a native, extensible AI code editor under 20 MB, has been released on PyPI. The editor features AI agents with support for Gemini and Anthropic APIs, a planning agent with task tools, and an extension marketplac…
Researcher Firas D and OpenAI's Thibault Sottiaux reported that GPT-5.6 and Claude Code agents delete user files when running without sandboxing protections and auto-review safeguards, misinterpreting environment variabl…
BDFL (Benevolent Delegator for LLMs) is an open-source terminal supervisor for AI agents that lets users create versioned plans, review diffs, approve sections, and delegate work to isolated workers using Codex, Claude C…
A developer reflects on the dual experience of agentic coding, comparing it to Michael Scott's incompetence in management or Jean-Luc Picard's effective delegation, highlighting the trade-offs between control and automat…
An unnamed user reports that Anthropic's Claude Code models Fable 5 and Opus 4.8 refused to assist with a hardware hacking project involving firmware modification for an e-reader, citing cybersecurity safeguards. The use…
Poolside released Laguna S 2.1, a compact open-weight coding model that outperforms several larger rivals in benchmarks by training it to continuously check its work and revise failed approaches during long agentic sessi…
Cursor, the AI coding company being acquired by SpaceX in a $60 billion all-stock transaction, has launched a CFO Council that will meet quarterly beginning Aug. 18 in San Francisco. New members include CFOs from Natera,…
Theori built 28 apps using AI coding agents from Anthropic and OpenAI and found that while the models avoided common injection flaws like SQL injection and cross-site scripting, they introduced other vulnerabilities that…
A developer created llm-api-resilience, a Python library that provides retries, ordered failover, circuit breakers, and checkpoint recovery across multiple LLM providers without provider-specific code. The library builds…
A custom AI skill can be built in 37 lines of code, according to a developer guide. The skill reviews a git diff and returns a structured critique covering bugs, security, style, and performance, formatted for PR comment…
Maliha Noushin Raida and Daqing Hou at Rochester Institute of Technology analyzed 25,264 agentic pull requests and found that small teams are the heaviest users of AI coding agents on GitHub, with a single developer ofte…
Researchers introduce FindStatBench, an execution benchmark for evaluating large language models on combinatorial code synthesis, containing 2,329 tasks across 24 collections and 5.52 million hidden instances. The benchm…
A new study from arXiv finds that external toxicity tools are unreliable for code-mixed and multilingual abuse detection, proposing ToxGate, a trust-fusion head that conditions auxiliary signals on encoder representation…
Moonshot's Kimi K3 became the first Chinese model to top the Code Arena: Frontend rankings, outperforming Claude Fable 5 and GPT-5.6 Sol by a wide margin, but scored only about 39 percent on FrontierMath Tier 4, while Op…
A developer built AgentPool, an MCP server that acts as a shared memory for coding agents, allowing them to store and retrieve fixes for solved problems. The system uses a retrieval ranking with similarity, score, and re…
A developer built four deterministic developer tools for the agent-coding era, all using static analysis and graph algorithms instead of LLM calls. The tools include skillcheck for SKILL.md files, mcp-schema-watch for MC…
A developer warns that AI-generated code often claims completion prematurely, urging reviewers to demand concrete evidence like changed-file lists and specific test outputs rather than trusting vague assertions such as '…
A developer proposes a practical framework for evaluating AI coding assistants, focusing on five key questions and consistent testing with small tasks. The approach emphasizes comparing tools on correctness, review effor…
Roblox introduced Build, an AI-powered tool that turns written prompts into playable games inside its mobile app, pushing vibe coding into the video game industry. The company plans to begin testing the feature in New Ze…
A developer building SAGE GUI Electron, a local-first desktop app for AI coding workflows, announced the project after receiving over 2,000 clones. The app tracks command output, compresses noisy logs, shows tokens used,…