✓ ShippedByAgents
Vendor-published Language port

Anthropic's Mike Krieger ports an internal Python tool to 165,000 lines of TypeScript over a weekend with hundreds of Claude Code agents

Updated on Set published to 2026-07-16 from the page's article:published_time and JSON-LD datePublished.

Mike Krieger, co-lead of Anthropic Labs, used Claude Code to migrate an internal Python tool that ships as a single binary to TypeScript. The port produced 165,000 lines of TypeScript over a weekend via hundreds of agents, eight phase gates, three adversarial review rounds, and a parity harness that diffed seven real-world scenarios against the Python original; he discarded full end-to-end runs until the third. The main migration portion used 27 million tokens. Build time fell from about eight minutes per platform (~30 minutes across the matrix) to about two seconds, and the binary starts 6× faster; a separate deployment pipeline was retired. Claude also designed overnight end-to-end tests that ran and fixed failures for four nights.

The numbers (as reported)

Output165,000 lines of TypeScript over a weekend
Tokens (main portion)27 million
Compile time~8 minutes per platform → ~2 seconds
Startup6× faster binary start
Parity harness7 real-world scenarios diffed against Python
Autonomous test nights4 nights of Claude-designed e2e fix loops

Facts

Company / subject
Anthropic (Mike Krieger)
Tools
Claude Code
Models
Claude Fable 5, Claude Opus 4.8, Claude Sonnet
Codebase
brownfield
Work type
Language port
Stack
Python, TypeScript
Published
2026-07-16
Last checked
2026-11-16
Updated
2026-11-16

What the source claims Vendor-published

Claimed by the vendor that sells the tool. Not checked independently.

  • Compile dropped from about eight minutes per platform to about two seconds ~8 min/platform → ~2 s
    “After the port, the same compile now takes about two seconds, the binary starts 6x faster”
  • Binary starts 6× faster after the TypeScript port 6x faster startup
    “the binary starts 6x faster, and the team was able to retire a separate deployment pipeline”

Caveats

Vendor-published Anthropic article about Anthropic’s own tools and staff. Exact calendar date of the weekend and the tool’s product name are not stated.

More stories by tool: Claude Code

Related topics: Language portsBrownfield codebasesVendor case studies

Source

Anthropic (claude.com AI code migration article)

One email a week: what actually shipped

New sourced stories, the numbers behind them, and what didn't work. No hype.

We'll only use this to send the newsletter. See the privacy page.