跳到主要内容
知仓学习社ZHICANG

background-agent-pings

Background Agent Pings

不碰外部(只输出文字)无严重或高危命中parcadei/Continuous-Claude-v3

它会碰到什么

扫了多少1 个文本文件,1 KB
它会碰到什么不碰外部(只输出文字)
命中总数0 处
命中统计严重 0 · 高 0 · 中 0 · 低 0

这一栏是扫描器报的事实,不是结论。命中多不等于有毒(安全工具、规则库、示例脚本本来就会包含危险写法),命中少也不等于干净。它和你手上的凭据、文件、网络有什么关系,需要你自己看。

技能内容

Background Agent Pings

Trust system reminders as agent progress notifications. Don't poll.

Pattern

When you launch a background agent, continue working on other tasks. The system will notify you via reminders when:

  • Agent makes progress: Agent <id> progress: X new tools used, Y new tokens
  • Agent writes output file (check the path you specified)

DO

1. Task(run_in_background=true, prompt="... Output to: .claude/cache/agents/<type>/output.md")
2. Continue with next task immediately
3. When system reminder shows agent activity, check if output file exists
4. Read output file only when agent signals completion

DON'T

# BAD: Polling wastes tokens and time
Task(run_in_background=true)
Bash("sleep 5 && ls ...")  # polling
Bash("tail /tmp/claude/.../tasks/<id>.output")  # polling
TaskOutput(task_id="...")  # floods context

Why This Matters

  • Polling burns tokens on repeated checks
  • TaskOutput floods main context with full agent transcript
  • System reminders are free - they're pushed to you automatically
  • Continue productive work while waiting

Source

  • This session: Realized polling for agent output wasted time when system reminders already provide progress updates

想直接用这个技能?

本站把开放许可(MIT / Apache 等)的技能按仓库打包整理到网盘,点一下转存到你自己的网盘,不用一个个从 GitHub 拉。许可未声明的技能只给原始仓库链接,不打包。