Pydantic AI adds Claude Code-style coding tools in one line, countering migration claims
samuelcolvin · x · 2026-11-04
After hearing at a conference that someone left Pydantic AI for the Claude Agent SDK for its coding tools, Pydantic team member Marcelo Trylesinski shows the same powers exist in Pydantic AI — in one line.
- Install with uv add "pydantic-ai-harness[coder,anthropic]", then add Coder() and LocalWorkspace('.') to any Agent.
- Coder() provides readfile, writefile, all-or-nothing editfile, ripgrep-based listfiles/grep (respecting .gitignore), shell (including long-running commands), and delegatetask for spawning subagents.
- It auto-loads AGENTS.md and CLAUDE.md, clears stale tool results when context fills, and enables prompt caching. Works with any model and typed outputs.
- In a demo on a project with a broken slugify(), the agent ran tests, read the failure, rewrote the function with a regex, re-ran tests, and flagged the edge case it chose not to handle.
The post includes a feature-by-feature mapping between the Claude Agent SDK and Pydantic AI equivalents.
More from coding & agent
- Weaviate ships Boost soft re-ranking in v1.38, with an 81k-product search recipe — CShorten30 · 2026-11-04
- AI can write more code, but engineers still must engineer the system — rseroter · 2026-11-04
- OpenAI acquires Pierre to turn Codex into a collaborative AI software-building platform — patience_cave · 2026-11-04
- Microsoft Clarity adds Agentic Analytics as AI agent traffic nearly doubles to ~2M in 3 weeks — OnlineInference · 2026-11-04
- Agentic Engineering vs AI Engineering: how the two disciplines actually differ — Pavan_Belagatti · 2026-11-04
- Grok Bot Books a Flight End-to-End: Search, Payment, and Bank Verification in One Chat — XFreeze · 2026-11-04