Anthropic deleted 80% of Claude Code's system prompt with no measurable loss. A four-layer anatomy of the agent harness, three tests for what should have been a script, and a worked example where a well-written skill dissolves into a git hook.
Claude Code's /loop command runs prompts on recurring intervals. Here's how to use it for cost awareness, CI monitoring, and deploy watching.
55K tokens vs 200. Perplexity dropped MCP. Here's what the benchmarks actually show about AI tool integration approaches.
Scaling laws are hitting diminishing returns, and the AI research community is pivoting. Here's the data on where things actually stand.
24 of 111 Claude Code sessions exceeded $50. The top cost $180 over 10 days. Here's the compounding cost problem and 5 strategies to fix it.
Claude Code v2.1.80 exposes rate limit data in the statusline JSON. Here's how to display it and stop getting surprised by throttling.
The --channels flag lets you route permission prompts to Telegram. Here's how to reclaim dangerous mode without losing control.