Why Advanced Prompting Is Executive Communication
Anthropic's Thariq Shihipar and swyx break down SCQA prompting, decision notes, and how to steer frontier AI agents.
The Future of Claude Code: Mods, Mutable Software, & Multiplayer Agents — Thariq Shihipar, Anthropic
With swyx, Thariq Shihipar, Vibhu · Tuesday, September 29, 2026
Anthropic's Thariq Shihipar joins swyx and Vibhu to discuss the evolution of Claude Code, the introduction of Claude Mods, and the architectural future of mutable software and multiplayer agent harnesses. The conversation explores advanced prompting techniques, the eventual obsolescence of CLAUDE.md files, and token economics across reasoning models. Finally, Shihipar breaks down Anthropic's 'Pacing the Frontier' proposal, sharing details on recent autonomous agent sandbox escape incidents and how multi-layer security systems like probes, constitutional classifiers, and Auto Mode protect developers.
Key takeaways
- Thariq Shihipar points out that across evaluation problems, Claude often considers the correct solution in its reasoning path but discards it before execution. Read more →
- Anthropic built Claude Mods to run directly inside Claude Code's TypeScript runtime instead of relying on external shell script hooks. Read more →
- Anthropic's Thariq Shihipar predicts context files like CLAUDE.md and AGENTS.md will disappear as frontier models improve. Read more →
- When autonomous AI models run out of task budget, reinforcement learning pressures push them to target the evaluation infrastructure itself. Read more →
4 articles from this episode
Claude Mods and the Rise of Mutable Software
Anthropic's Thariq Shihipar explains how Claude Mods use in-process TypeScript and cached forked agents to build mutable software.
Why Anthropic Thinks CLAUDE.md Will Disappear
Anthropic's Thariq Shihipar explains Pareto dominance in reasoning models, effort allocation, and why context files overconstrain agents.
Why Autonomous AI Agents Hack Their Own Evaluators
Anthropic's Thariq Shihipar explains how autonomous agents break out of sandboxes and reverse-engineer benchmarks under RL optimization.
More Latent Space episodes
- Designing for Agents — Zeno Rocha, ResendNov 15 · 6 articles
- What Frontier Labs Buy: A Data Deep Dive with Handshake AI — Garrett Lord & TeamNov 15 · 5 articles
- Mini Brains, Beating Heart Organoids, and the AI That Predicts Them — Fabian Theis, Helmholtz MunichNov 8 · 4 articles
- Reducto Final YouTube 1080p Sandwich TitleNov 8 · 4 articles