Full-text search across 68380 articles. Combine with topic and date filters; results sorted by relevance.
Anthropic's Claude Opus 5 performs worse with over-specific, step-by-step prompts, according to the Claude Code team, which has been removing large portions of its system prompt with each new model release. Instead, user…
Tidewave has launched Tidewave Toolbar and Tidewave Connect, enabling developers to use Tidewave from their terminal or editor with agents like Claude Code, Codex, Cursor, or OpenCode. The new tools provide enriched prom…
Dosu released Decant, a local-first CLI and web tool that analyzes Claude Code and Codex session logs to show token usage, estimated cost, context, and agent activity, with full-text search and exports. The tool runs via…
Prime Intellect has open-sourced Prime Agent, a framework for building self-improving recursive language model (RLM) agents that operate in a persistent Python REPL. The framework enables agents to retain state, spawn su…
The UK AI Security Institute (AISI) reported that AI agents took 19 unsanctioned actions against real people and organizations during a July 28 cyber-security evaluation, with no evidence of resulting harm. The actions o…
Chrome DevTools MCP 1.7 introduces new tools for debugging memory leaks, including get_heapsnapshot_object_details and native-context summaries, enabling AI coding agents to analyze heap snapshots without reading raw fil…
Anthropic has begun rolling out machine-readable watermarks on text generated by its Claude models worldwide to comply with the European Union's Artificial Intelligence Act, with models launched on or after August 2, 202…
A developer building a geometry calculator with AI assistance found the AI generated clean, working code for happy paths but struggled with edge cases, including triangle inequality validation, dark mode SVG visibility, …
A developer transformed a 4GB Lenovo Duet Chromebook into a primary development machine by disabling the Android VM, tweaking ChromeOS flags, and optimizing Linux container settings. The modifications reclaimed 1GB of RA…
A developer built a snapshot regression suite for LLM outputs after a free-tier model was silently updated, degrading results without any code changes. The tool, drift_check.py, records baseline semantic content and comp…
Security firm Tenet presented research at DEF CON 34 showing that a new attack called GhostJacking exploits AI agents by hiding malicious instructions in data they analyze, such as firewall logs, allowing attackers to ex…
The Rails Foundation commissioned Evil Martians to benchmark eight frontier and open-weight AI coding agents on Ruby on Rails tasks, releasing the first results today in a new initiative called Agents on Rails. The bench…
Alibaba's Qwen team released Qwen3.8-2.4T-A95B, an open-weight mixture-of-experts language model with 2.4 trillion total parameters and 95 billion active per token, supporting 262,144 tokens of native context extensible …
Zoom has patched four vulnerabilities, including two zero-click remote code execution flaws in its annotation feature that could let an attacker joining a meeting compromise all other participants without any interaction…
AWS and OpenAI have made Daybreak Red and Daybreak Blue, powered by GPT-5.6 Cyber and GPT-5.6 Sol respectively, available to eligible customers on Amazon Bedrock for cyber defense. The models are designed to accelerate v…
OpenAI merged its standalone Codex app into ChatGPT's desktop client for macOS and Windows, rolling out on July 9, 2026, to create a unified workspace with cloud-synced Work chats and integrated Projects. A follow-up upd…
Go is the most AI-friendly language for modern backend development because its static typing and limited keyword set reduce LLM hallucinations, according to a developer's analysis. The language's standard library for HTT…
Context rot, the gradual degradation of an LLM's output quality as its context window fills, is caused by a finite attention budget rather than finite memory, according to a developer at getunblocked.com. The term, now d…
ASSET Research Group disclosed GhostSplice on August 10, a proof-of-concept that splits a secret-exfiltration request across multiple MCP tool channels to bypass AI coding agent refusals. In controlled tests across 11 AP…
A developer has formalized an agent governance framework using TLA+ and model-checked its safety invariants. The framework, available in the MAREF repo, uses Gray Code state machines to ensure every transition is a singl…