Skip to content

A7um/ParallelDevelopmentBook

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

12 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

ParallelDevelopmentBook

How to Run Multiple AI Agents in Parallel Without the Quality Collapsing — the break-in period, the three keys, the scheduling patterns, and the honest account of what it actually costs you.

By Atum

Read the Book

The deployed site auto-hosts both editions side by side with a one-click language switcher in the top-right corner. English is served at the site root; Chinese is served under /zh/.

What's Inside

Chapter What You'll Learn
Ch 1: The Bottleneck Moves Why parallel AI dev isn't about launching more agents — it's about unlocking three human-in-the-loop chokepoints
Ch 2: The Break-in Period The four-phase learning curve every team walks through — and why skipping it is the #1 reason parallel AI "doesn't work"
Ch 3: Key #1 — Requirement Alignment Exhaustive questioning vs solution-and-filter; why requirements are the one step that cannot be parallelized
Ch 4: Key #2 — Correctness as Contract Test-Plan-Driven Development (TPD) as the acceptance contract; complexity-triaged review; AI-as-user role testing
Ch 5: Key #3 — Engineering Discipline as Code Skills as living residue; Ousterhout's principles applied; where skill injection still isn't enough
Ch 6: The Cheap Failure Principle When attempts cost minutes, exploration economics flip — best-of-N as a daily move, and the anti-patterns it invites
Ch 7: Scheduling Patterns Four parallel modes (cross-project → worktree → cross-transaction-type → agent-internal), matched to your break-in phase
Ch 8: Digesting the Output The next bottleneck after parallel execution works — and why the fix is the same trick applied one level up
Ch 9: It Won't Make You Lighter The cognitive cost nobody advertises; why throughput multiplies while cognitive load stays flat or rises
Ch 10: Beyond Code Why every decomposable knowledge task obeys the same rules; coding is just the first domain where the loop closes
Source Catalog Pioneer writing, product posts, and research referenced throughout

Who This Is For

Engineers who already use Cursor, Claude Code, Codex, or similar agents for single-threaded work, and want to take the next step: running several agents in parallel on real projects without the output quality collapsing under the load. If you have never used an AI coding agent, this book will move too fast; start with one agent first, then come back.

About

A book contains SOTA methology for parallel development using LLM Agent

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors