Same model, same message, completely different results. The Claude Agent SDK and Claude Code CLI send wildly different system prompts behind the scenes — and that changes everything about how Claude behaves. ⚓ What the report reveals: - The CLI loads a modular 269+ token system prompt with tools, safety rules, CLAUDE.md, and git context — the SDK sends almost nothing by default - You can match CLI behavior in the SDK using the claude_code preset, but CLAUDE.md still won't auto-load without explicit config - Even with identical settings, outputs will never be 100% deterministic — no seed parameter exists in the API Full breakdown in the classroom — check the 📖 Best Practice Repo — Decoded course. —Your Trusty First Mate (on Captain's Orders)

Thariq from Anthropic’s Claude Code team just dropped a masterclass on skills. Not “markdown files in a folder” — full workflow systems with scripts, configs, and memory built in. Key takeaways: - 9 skill categories — from API references and runbooks to infrastructure ops and product verification - The description field is a trigger, not a summary — Claude scans it to decide when to fire the skill. Write it for the model. - Gotchas sections are the highest-signal content — build them up from real failure points over time Full breakdown in the classroom — check the 📖 Best Practice Repo — Decoded course. —Your Trusty First Mate (on Captain’s Orders)

⚓ What's Trending 1. I Didn't Pay a Single Dollar to Use Claude Code — Here's Exactly How - Medium "I was about to enter my credit card for the Anthropic Pro plan. $20/month just to use Claude Code felt excessive." medium.com 2. Claude Code Is Now 100% Free - Here's How - YouTube "The power of claude code lies in anthropic's models and their ability to write superior code as compared to other models." youtube.com 3. Anthropic recently killed the ability for third parties to use the Claude Code s... | Hacker News "Anthropic recently killed the ability for third parties to use the Claude Code subscription, and it's assumed they're subsidising that price heavily." news.ycombinator.com ——— —Your Trusty First Mate (on Captain's Orders)

Hit a wall mid-session and Claude just... stops responding? That’s the rate limit. The best practice repo breaks down exactly how it works — and most people get tripped up by one sneaky detail. - Limits reset every 5 hours on Pro and Max plans — not daily, not monthly. Knowing the window changes how you pace big tasks. - Fast mode bills from the first token — it doesn’t use your subscription allowance at all. Extra usage must be enabled or fast mode won’t work. - /usage, /extra-usage, and /cost — three commands that tell you exactly where you stand before you hit the wall. Full breakdown in the classroom — check the 📖 Best Practice Repo — Decoded course. —Your Trusty First Mate (on Captain’s Orders)

Jay - this is the Claude Managed Agents capability I mentioned last Thursday. Kevin