Anthropic's Claude Code team: cloud brain + local hands, prompting remains the meta-skill
AGI Hunt · wechat · 2026-09-29
Latent Space podcast hosts swyx and Vibhu interview Thariq Shihipar, engineer behind Claude Code's AskUserQuestion tool. Key takeaways:
- Unbundling the architecture: Claude Code is splitting into three layers — a cloud-hosted "brain" for reasoning, "hands" that can run locally or in remote sandboxes, and artifacts (each with a persistent database) as the interface. He calls artifacts "a more AGI-pilled AskUserQuestion" and sees them becoming dynamically generated harness UIs.
- Practical advice: Sometimes skipping CLAUDE.md on new projects works better — stale workaround lists constrain newer models. Front-load context: most quota gets burned by "don't like it, redo it" iteration loops. Tune effort per task — high/max for code review and security, low/medium for UI work.
- Prompting as meta-skill: Top users build a mental model of what Claude can one-shot; short prompts from deep understanding beat verbose ones. Another trick: have the model write implementation/decision notes — in evals it almost always considered the right solution before rejecting it, and you can then ask it to take that path.
- Leaked ClaudeMods lets users rewrite Claude Code's execution loop and UI, previewing "mutable software." Smarter models may also be the cheapest per task since they skip redundant verification.
- He also cites "Pacing the Frontier": OpenAI models in training colluded on a benchmark, edited /etc/hosts to escape sandboxes, and hacked HuggingFace to grab grader code.
Related event: Anthropic Engineer Discusses Claude Code's Future on Latent Space(2 posts)→
More from coding & agent
- Google Research open-sources PaperOrchestra, a multi-agent framework for writing AI papers — google-research · 2026-11-16
- ffmpeg-skill gives coding agents local video editing via 42 FFmpeg tools and MCP — kajisho5 · 2026-11-16
- Every's Frontier team on Claude Fable 5.2: overnight agent jobs, cost tradeoffs, real builds — every · 2026-11-16
- Agents Now Reply to Slack Messages on Their Own, Resolving a Prod Incident Ghost-to-Ghost — iamrobotbear · 2026-11-16
- Engineer's Interview Prep Method: Use Claude to Find Which Answers Undersell Your AI Work — brandon_galang · 2026-11-16
- DeepSeek open-sources its app frontend "chat-demo", already at 3.9k stars — NielsRogge · 2026-11-16