A hook should be a function call, not a shell

说明与用途

在 Claude Code 中，每个钩子都作为单独的 shell 脚本运行，产生多个进程并导致延迟。通过将钩子移动到单个插件模块并在现有进程中执行，执行时间降至接近零，提升了性能。

问题

Each hook in Claude Code ran as a separate shell script, spawning multiple processes and causing significant latency.

结果

Hook execution time dropped from ~90 ms to near zero, reducing overall latency and resource usage.

工作流

• Identify hooks that can be converted to function calls
• Create register.js to load hooks once per session
• Move hook logic into hooks/mods/*.js as plain functions
• Update workkit to use the new module for hook execution
• Test with claude plugin test to ensure no Claude Code dependencies

证据

Until this week every hook was a shell script. Each fire went through a loader script, which started bash, which usually called jq to read the event, and often node for the real logic.

来源

Claude Code can now load a plugin module once per session and call it on each event.

来源

workkit 0.79 moved thirteen hooks into one such module, and fifteen live there now.

来源

A bare bash, jq and node chain alone took about 38 ms. A function call inside a process that is already running rounds to zero.

来源