Registrieren

Registierung erfolgt in Kürze...
Fleebs-Logo
Details werden geladen...

A Claude Code skill was eating 200,000 tokens before answering a single question - The New Stack

Anthropic's /claude-api skill was silently consuming 200K tokens per request. A new Claude Code update cuts that to 25K by loading docs on demand.

Ähnliche Seiten

https://thenewstack.io/claude-fable-cost-model-triage/

Claude Fable cost $9 in one coding test. GPT-5.5 cost $1.50. Model triage is the new AI skill. - The New Stack

https://thenewstack.io/claude-fable-cost-model-triage/
https://thenewstack.io/caveman-mode-token-savings/

Getting Claude Code to grunt in Caveman-speak might not save as many tokens as you think - The New Stack

https://thenewstack.io/caveman-mode-token-savings/
https://thenewstack.io/anthropic-claude-design-overhaul/

Anthropic overhauled Claude Design to fix the handoff. A designer and an engineer disagree on whether it worked. - The New Stack

https://thenewstack.io/anthropic-claude-design-overhaul/
https://thenewstack.io/agentjacking-sentry-mcp-attack/

A public Sentry key is all it takes to hijack Claude Code, Cursor, and Codex - The New Stack

https://thenewstack.io/agentjacking-sentry-mcp-attack/