AI agents: when artificial intelligence takes action
AI agents are the big shift right now: models that don't just answer, but use tools, execute tasks and act autonomously. Here's the dossier and the latest news.
Latest AI agent news
- Claude Code 2.1.153 — Fixed subagent (Agent tool) frontmatter MCP servers ignoring --strict-mcp-config, --bare, remote mode, enterprise managed MCP co — Claude Code
- Claude Code 2.1.162 — Fixed MCP per-server timeout config values below 1000 ms being floored to a 1-second watchdog that aborted every tool call; sub- — Claude Code
- Claude Code 2.1.7 — Enabled MCP tool search auto mode by default for all users. When MCP tool descriptions exceed 10% of the context window, they are — Claude Code
- Claude Code 2.1.202 — Added a "Dynamic workflow size" setting in /config for controlling how large Claude generally makes dynamic workflows (small/med — Claude Code
- Claude Code 2.1.205 — Improved agent view: rows now show a colored state word and a classifier-written headline instead of raw tool call text, and the — Claude Code
- Claude Code 2.1.187 — Fixed --json-schema and workflow agent({schema}) structured output: the model can no longer re-call StructuredOutput indefinitel — Claude Code
- Claude Code 2.1.114 — Fixed a crash in the permission dialog when an agent teams teammate requested tool permission — Claude Code
- Claude Code 2.1.91 — Added MCP tool result persistence override via _meta["anthropic/maxResultSizeChars"] annotation (up to 500K), allowing larger res — Claude Code
- Claude Code 2.1.9 — Added auto:N syntax for configuring the MCP tool search auto-enable threshold, where N is the context window percentage (0-100) — Claude Code
- We've launched Claude Platform on AWS, bringing the Claude API to Anthropic-managed infrastructure accessible through AWS, with AWS billing and IAM au — Anthropic (release notes)
What is an AI agent?
An AI agent combines a language model, tools (search, code, APIs) and a decision loop to accomplish a multi-step goal with minimal human intervention. It's the jump from 'assistant' to 'doer'.
Claude Code, MCP and agentics
Claude Code illustrates agentics applied to development. The MCP protocol standardizes connecting agents to tools and data. Together they sketch an infrastructure of interoperable agents.
Promise and guardrails
Autonomy multiplies value but also risk: irreversible actions, cascading errors, security. Good deployments keep a human in the loop and limit permissions.
Frequently asked questions
What is an AI agent?
A system combining a model, tools and a decision loop to accomplish a multi-step task autonomously.
Is Claude Code an agent?
Yes, it's a developer agent: it reads code, runs commands, edits files and runs tests.
What role does MCP play for agents?
The Model Context Protocol standardizes connecting agents to external tools and data; see our dedicated page.
Claude News is published by Héra SASU. Independent media, not affiliated with Anthropic.