AI AGENTS

A July 2026 paper from Tencent Hunyuan turns agent harnesses into behavior-level maps. The useful lesson for builders is simple: code search is not enough when one behavior spans prompts, tools, state, permissions, and runtime policy.

SkillHone is a July 2026 paper about evolving agent skills across sessions. The useful takeaway for developers is simple: do not save only the latest SKILL.md. Save the decisions that explain why it changed.

How to set up Entire's regional Git mirrors for AI coding agents. Covers installation, mirroring, integrations with Claude Code, Codex, Cursor, and Factory AI.

StateM pushes Terminal-Bench 2.1 to 95.3% raw accuracy by scaling the harness around the model. The lesson for coding-agent teams is that runbooks, state, and recovery loops now matter as much as model choice.

Open-source tool gives Claude Code, Codex, and other agents their own isolated Linux VM on your machine - network firewall included, no cloud account required.

A deep dive into the agent orchestration behind the Bun Rust rewrite - the workflow architecture, adversarial review gates, what one human actually did, and the Zig vs Rust debate including Andrew Kelley's response.

A companion guide to the Composio CLI video: one command-line layer that lets Claude Code, OpenClaw, Codex, and other agent harnesses search, authenticate, and execute tools across 1,000+ apps.

ByteDance's Dockerless paper asks whether coding-agent patches can be verified without spinning up per-repo environments. The practical answer is not replace CI. It is use cheaper evidence before CI.

The architecture side of loop engineering: plan/act/verify cycles, convergence criteria, retry policies, budget-bounded loops, and the loop-until-dry pattern. Concrete TypeScript-shaped patterns for building agent loops that stop when they should.

A new Vera paper tests Codex, Claude Code, OpenClaw, and Hermes with executable safety cases. The useful lesson is not panic. It is evidence-grounded agent QA.

A fair comparison of Braintrust, Langfuse evals, Promptfoo, DeepEval, Ragas, and OpenAI Evals: offline vs online evals, LLM-as-judge, CI integration, and dataset management for agent testing.

A new benchmark shows GLM 5.2 processing 59 transactions and producing VAT returns off by only 7 pence - at $2.73 versus typical accounting fees of $1,000+. Here is what the benchmark actually tested, where the model failed, and why the HN discussion focused on liability.

A fair, sourced comparison of the three LLM observability platforms teams reach for once agents hit production: Langfuse's open-source tracing and prompt management, Braintrust's eval-first workflow for regressions, and Helicone's drop-in proxy for logging and cost control. Architecture, pricing model, self-hosting, and which to pick by workload.

A fair, sourced comparison of the four runtimes developers reach for when they want a coding agent talking to a model on their own hardware instead of an API: Ollama's convenience, LM Studio's GUI, vLLM's throughput, and llama.cpp's control. What each is actually for, and which to pick.

Meta launches Muse Spark 1.1 through the new Meta Model API - a 1M-token-context model for personal agentic tasks with OpenAI-compatible endpoints, $20 free credits, and pricing that undercuts the competition.

pgvector, Pinecone, Qdrant, Weaviate, Chroma, Milvus, and Turbopuffer compared on hosting model, filtering, scale, and cost for RAG.

Security researchers discovered a prompt injection vulnerability in GitHub's Agentic Workflows that allows attackers to extract private repository contents through public issues.

Lilian Weng argues self-improving AI won't start with models rewriting their weights - it starts with the harness. Here's what that means for developers building agents.

Microsoft Execution Containers (MXC) give your AI agents policy-driven sandboxing across Windows, Linux, and macOS. TypeScript SDK, JSON config, multiple isolation backends. Here is how to use it.

Claude Code and Codex both ship great agents and terrible transcripts. AgentCanvas is a visual adapter that puts the artifacts, decisions, and handoffs on one board so the next agent and the next human can see them.