Convert a codebase to TypeScript with a swarm of cheap agents
A shared Claude Code recipe uses Sonnet 5.5 as the lead and a swarm of Haiku 5.5 agents that each convert one file to TypeScript. The hardest files go to Sonnet, which then merges every diff and runs the test suite.
Source
@0X_Vetra on X · 2026-10-17
Written in our own words. Credit to the original author. Link checked 2026-10-22.
What they did
A shared Claude Code recipe uses Sonnet 5.5 as the lead and a swarm of Haiku 5.5 agents that each convert one file to TypeScript. The hardest files go to Sonnet, which then merges every diff and runs the test suite.
Result (as claimed)
24 files converted in 3 minutes for $0.71 with 148 of 148 tests passing, while Opus alone was 22% done.
Tools they used
Cost: Claude Code: from $17/mo; Claude Sonnet 5.5: from $2 in / $10 out per 1M tokens; Claude Haiku 5.5: from $0.10 in / $0.50 out per 1M tokens.
Set it up
Pick a contained repo or task
Start on a branch or side project where mistakes are cheap.
Install and configure
Set up Claude Code, Claude Sonnet 5.5, Claude Haiku 5.5, TypeScript. Read what it can run on your machine and limit permissions where possible.
Run it on a real task
Give it a well-scoped issue and let it work. Review the diff like any pull request.
Add guardrails
Keep tests and CI in the loop, and never let an agent push to main or touch production secrets unattended.
Copy the full setup prompt
Paste into your agent and it will walk you through the setup, asking before anything that sends, posts or spends.
You are helping me set up this AI use case end to end in Claude Code. Goal: Convert a codebase to TypeScript with a swarm of cheap agents What the original author did: A shared Claude Code recipe uses Sonnet 5.5 as the lead and a swarm of Haiku 5.5 agents that each convert one file to TypeScript. The hardest files go to Sonnet, which then merges every diff and runs the test suite. Reported result (as claimed by the author): 24 files converted in 3 minutes for $0.71 with 148 of 148 tests passing, while Opus alone was 22% done. Source: https://x.com/0X_Vetra/status/2111385535511949502 Tools they used: Claude Code, Claude Sonnet 5.5, Claude Haiku 5.5, TypeScript Category: Dev & ops Please: 1. Ask me 3 short questions about my situation (accounts I have, budget, what I want the output to look like). 2. Propose a step-by-step plan using these tools or cheaper alternatives, with the time and cost of each step. Use official pricing pages; say "check pricing" if unsure. 3. Then do the setup with me one step at a time, waiting for my OK before anything that sends messages, posts publicly, spends money or deletes data. 4. Finish with a checklist to run it again next time.
You are helping me set up this AI use case end to end in Codex. Goal: Convert a codebase to TypeScript with a swarm of cheap agents What the original author did: A shared Claude Code recipe uses Sonnet 5.5 as the lead and a swarm of Haiku 5.5 agents that each convert one file to TypeScript. The hardest files go to Sonnet, which then merges every diff and runs the test suite. Reported result (as claimed by the author): 24 files converted in 3 minutes for $0.71 with 148 of 148 tests passing, while Opus alone was 22% done. Source: https://x.com/0X_Vetra/status/2111385535511949502 Tools they used: Claude Code, Claude Sonnet 5.5, Claude Haiku 5.5, TypeScript Category: Dev & ops Please: 1. Ask me 3 short questions about my situation (accounts I have, budget, what I want the output to look like). 2. Propose a step-by-step plan using these tools or cheaper alternatives, with the time and cost of each step. Use official pricing pages; say "check pricing" if unsure. 3. Then do the setup with me one step at a time, waiting for my OK before anything that sends messages, posts publicly, spends money or deletes data. 4. Finish with a checklist to run it again next time.
You are helping me set up this AI use case end to end in Grok Bot. Goal: Convert a codebase to TypeScript with a swarm of cheap agents What the original author did: A shared Claude Code recipe uses Sonnet 5.5 as the lead and a swarm of Haiku 5.5 agents that each convert one file to TypeScript. The hardest files go to Sonnet, which then merges every diff and runs the test suite. Reported result (as claimed by the author): 24 files converted in 3 minutes for $0.71 with 148 of 148 tests passing, while Opus alone was 22% done. Source: https://x.com/0X_Vetra/status/2111385535511949502 Tools they used: Claude Code, Claude Sonnet 5.5, Claude Haiku 5.5, TypeScript Category: Dev & ops Please: 1. Ask me 3 short questions about my situation (accounts I have, budget, what I want the output to look like). 2. Propose a step-by-step plan using these tools or cheaper alternatives, with the time and cost of each step. Use official pricing pages; say "check pricing" if unsure. 3. Then do the setup with me one step at a time, waiting for my OK before anything that sends messages, posts publicly, spends money or deletes data. 4. Finish with a checklist to run it again next time.
The Daily 5
Get five new use cases tomorrow morning
Free. Real sources. Unsubscribe any time.
Guides for this use case
How to run multiple coding agents together · How to vibe code an app with Claude Code · How to port code to another language with AI
Related use cases
Benchmark your agent harness against Claude Code with hidden judges
The founders of Relay, a local-first agent supervisor, pre-registered a 5-test suite with hidden judges and ran their tool, Claude Code, Prime Agent and Hermes through the same token meter on Sonnet 5.5 and Haiku 5.5, then published the losses along with the wins.
Port a C server to Rust with one prompt and a test loop
Claude Code's creator handed Sonnet 5.5 a whole C server and the tests, then let it loop: do the job, run the tests, fix, run again. The same port with the top model took 12 hours and about twice the cost.
Run cheap DeepSeek Flash workers inside Claude Code with Opus as the boss
A builder plugged DeepSeek Flash into Claude Code as the worker model and kept Opus 5.5 as the lead that plans and checks. He cites DeepSeek's own chart where Flash scored 77.0 vs 74.0 on the same agent benchmark.
Make a short film about vibe coding from real Claude Code sessions
A creator made Slow Me Down, a short film about a dad who vibe codes a game for his daughter and makes his own character weaker so she always wins, until she asks Claude why. It was made with Sonnet 5.5 using real Claude Code sessions.