Thariq Shihipar

Anthropic launches Claude Agent SDK, a generalized agentic framework evolved from Claude Code that enables developers to build versatile AI agents with computer access and tool use capabilities.

“Author of "Anthropic Launches Claude Agent SDK for Building Versatile AI Agents" in Anthropic”

“Co-creator of Claude Code who announced departure before Anthropic's IPO filing, saying he was taking a break before diving into something new.”

“Anthropic product leader discussing Claude Code evolution, Claude Mods, Claude Tag, Projects, and agent security challenges.”

“Left Anthropic ahead of IPO, was one of Claude Code's most visible builders.”

“Anthropic team member who deleted 80% of Claude Code's system prompt as models absorbed harness capabilities.”

“Angel investor in Discovered Materials seed round for AI-powered materials discovery.”

“Created guide on updated prompting best practices for Anthropic models”

“Authored a guide on updated prompting best practices for Claude models, explaining changes in context engineering and model capabilities.”

“Authored Anthropic's guide on context engineering best practices for Claude 5 generation models, describing how to optimize prompts and system guidance.”

“Discussed Claude Fable model at AIEWF keynote, describing how models are grown rather than designed.”

“Prototyping video games with Claude, discussing bottlenecks in AI capability versus human blind spots.”

“Anthropic executive who explained experiment in Claude Code to identify Chinese users and prevent unauthorized resellers and model distillation.”

“Claude Code team member participating in a fireside chat about prompting strategies and AI model capabilities.”

“Anthropic spokesperson who clarified that identity verification applies only to a small subset of flagged users as part of appeals process.”

“Published an X post arguing that AI agents should produce single-file HTML instead when creating files, which gained 4.4 million views in 16 hours.”