Claude Code Adds Self-Diagnostic Command
新智元 · wechat · 2026-07-11
Anthropic released version 2.1.205 for Claude Code, with the most notable addition being the /checkup command. It automatically checks and organizes project skills, MCPs, plugins, and CLAUDE.md configurations, clears unused content, splits scattered rules into more maintainable structures, handles slow hooks, enables auto mode by default, and pre-approves common read-only commands.
The article emphasized that the core value of these capabilities isn't just convenience, but the automated governance of context—the scarcest resource in Agent scenarios. This reduces the window space taken up by useless configs and lowers the probability of model forgetting, drifting, or slowing down. The author likened it to "garbage collection for the context world."
The new version also introduces stronger security guardrails: prohibiting tampering with session transcripts and forcing confirmation before executing rm -rf with unresolved variables, preventing catastrophic accidental deletions caused by excessive Agent permissions. The overall conclusion is that Claude Code is evolving from a "tool requiring manual human oversight" into an "autonomous agent capable of self-diagnosis and self-organization."
More from coding & agent
- Soft Clamp cuts tool-call overuse in multi-teacher distillation, from 13.7% to 9.0% — antgroup · 2026-07-21
- Agent harness memory loss and compaction are still a major usability problem — adityaag · 2026-07-21
- SpecJudge runs locally on Ollama to pick the right-sized AI model for your project — jokiruiz · 2026-07-21
- A developer maps out six design rules for CLIs that humans and AI agents can both use — yujiezha · 2026-07-21
- A coding-agent skill that forces ADHD-friendly, answer-first output — ayghri · 2026-07-21
- A set of agent skills for CAD, robotics, and hardware design — earthtojake · 2026-07-21