Category: tech > ai > agents

Claude Code's /goal command runs an autonomous loop: take an action, check the goal condition, repeat until met, replacing step-by-step prompting.

Reflexion lets an LLM agent learn from failures without fine-tuning, writing a verbal post-mortem and feeding it back as context, lifting GPT-4 to 91% on Hum...

A short clip claims Claude Code now auto-generates a Live Artifact summarizing code and PRs at the end of a session to share with teammates.

A short teaser for loop engineering, the idea that AI coding agents work best run in an iterative loop rather than expecting one-shot results.

Open-source MIT orchestrator that drives coding agents through a structured plan, running each task in a fresh session plus multi-agent code review.

Loop engineering is an agentic workflow where you set one goal and let agents self-prompt through discovery, planning, parallel execution, and verification.

Claude Code Dynamic Workflows runs hundreds of parallel sub-agents in one session, compressing multi-day coding into a day at much higher token cost.

Claude Code Workflows lets a workflow.js script invoke subagents in loops, sequences, or parallel, with structured output schemas to chain agents reliably.

Dozens of undocumented Claude Code config fields hidden in the npm source: command-rewriting hooks, persistent agent memory, and a plain-English auto-approva...

A single-binary Go coding agent that runs with no runtime dependencies. Provider-agnostic with intentional minimalism: four core tools and an extension model.

An MIT-licensed, terminal-first AI coding agent built only for DeepSeek, engineered around its prefix-cache to keep token costs low across long sessions.

Community directory cataloguing every connector in Anthropic's official Claude Connectors Directory, 414 verified MCP servers across 30 categories.

Google's autonomous coding agent takes a GitHub repo and a prompt, plans with Gemini 3 Pro, then produces a reviewable diff and opens a PR.

AI lowers the bar to build software, but engineers shift from writing code to managing agent teams and owning quality. Junior hiring is down nearly 20% since...

A Google Cloud demo shows one person playing five roles to build and deploy a full feedback app using Claude Code with Claude models on Vertex AI.

Perplexity open-sourced Bumblebee, a read-only scanner that continuously checks developer machines for risky packages, extensions, and AI tool configs.

A teaser for a video on building AI agents with memory across sessions. It lists topics like memory stores and prompt caching without showing how.

Three rules for shipping production agents: build them only for ambiguous high-value tasks, keep the architecture minimal, and inspect the context window.

Claude Code computer use can drive native desktop apps like Final Cut Pro, QuickBooks, and Lightroom, automating repetitive workflows for knowledge workers.

Anthropic's MCP tunnels let a managed agent act on data living on your own server through an MCP connection, so sensitive data never leaves your infrastructure.