Real Skill packageSource verifiedClawHub registry

subagent-first

AI 助手不是不够聪明，是被重活占住了：你让它整理 40 个文件，它一头扎进去十几分钟不抬头——你想问进度、想补需求、想改方向，只能干等。这套作业纪律解决的就是这件事：凡是预计要跑多轮的活（多文件/长研究/批量/长搜索），一律派给后台子代理并行执行，主代理只保留四个动作——拆解、派活、汇总、对话，永远保留随时回你话的能力。我们自己在多智能体团队里天天跑，踩过的 12 个坑全部写进正文：子代理自报成功却发现产出是空的、context 没写全导致跑偏返工、两个子代理并行写同一文件互相覆盖、把需要用户拍板的活派给问不了人的子代理、跨会话长任务错用子代理导致某天悄悄停掉……① 决策树四问判定轻重（一轮工具调用能做完吗/需要用户拍板吗/要活过本次会话吗/能拆成互不依赖的块吗）② 派活四要素（goal / context 必须自包含——子代理看不到你和用户的对话 / output_schema / 验收方式）③ 五步流程：拆解→派活→并行调度→验收→收口 ④ 验收清单：存在性/合规性/真实性/完整性四组逐项打勾，并随机抽一处回一手来源核对 ⑤ 6 种委派模式 + 12 条反模式。配套 scripts/delegation_planne…

Identity and source

Publisher attributionlingyu9495-sourceregistry owner unverified by skillvetai
Functional categoryAgent Engineering, Security & Governanceautomatically inferred · 55% rule confidence
Package forminstruction with code9 recorded files
Canonical sourceClawHub registryclawhub:lingyu9495-source:subagent-first
Open canonical source ↗

Platform declarations

These states come from the source or distribution context. None of the entries below are SkillVetAI compatibility test results.

OpenClawnative officialProvenance: registry distribution

Independent structural checks

These checks parse the fixed package against dated platform rules. They do not execute the Skill or verify task behavior.

Claude Codepasses structure
Checker 0.1.0 · agent-skills-2026-08-13+claude-code-docs-2026-08-13 · 9/26/2026.claude/skills/subagent-first

Runtime, accounts, dependencies, permissions, network behavior and task quality remain untested.

OpenAI Codexpasses structure
Checker 0.1.0 · agent-skills-2026-08-13+codex-docs-2026-08-13 · 9/26/2026.agents/skills/subagent-first

Runtime, accounts, dependencies, permissions, network behavior and task quality remain untested.

OpenClawpasses structure
Checker 0.1.0 · agent-skills-2026-08-13+openclaw-docs-2026-08-13 · 9/26/2026skills/subagent-first

Runtime, accounts, dependencies, permissions, network behavior and task quality remain untested.

Installation and inspection

This command is recorded from the source ecosystem and resolves the registry's latest release. The fixed release shown on this page should be inspected before adoption.

clawhub install @lingyu9495-source/subagent-first
clawhub inspect @lingyu9495-source/subagent-first --version 1.0.0

Security evidence

SkillVetAI static result: no findings detected

This automated, non-executing scan is bound to this release hash. It is not a safety certification and may contain false positives or false negatives.

Status
completed
Coverage
full text content
Files
9 / 9 inspected as text
Checked
9/26/2026, 4:55:18 AM
Scanner
0.1.3
Policy
1.0.3
2 inferred permission indicators
  • network access — automatically inferred
  • filesystem read — automatically inferred
3 dependency and API indicators
  • api: clawhub.ai
  • api: jinli-vault-1372591613.cos.ap-guangzhou.myqcloud.com
  • api: skillpay.alipay.com
External clawhub result: clean

This is registry-supplied evidence for the recorded release, not an independent SkillVetAI scan. Check the canonical source for the full report, scanner versions, scope, and current moderation state.

Evidence checked
9/25/2026, 9:43:28 PM
Release binding
Matches this record
  • vt: clean
  • skillspector: suspicious
  • llm: clean

Recorded files

The catalog stores hashes and an inventory summary for change detection. It does not republish the package contents.

Package content hashsha256:bba4dcdcc8527b30cf7dcfdcda984571c8be67739e471eca0a4587bb0aaf06b9
Show up to 9 recorded paths
  • references/hermes-runtime-reality.md
  • references/委派模式库.md
  • references/案例与致谢.md
  • scripts/delegation_planner.py
  • skill-card.md
  • SKILL.md
  • templates/子代理验收清单.md
  • templates/派活工单.md
  • 使用说明_README.md

Source changelog

subagent-first 1.0.0 - Initial public release of the "子代理优先 · 主代理不下场" discipline for agent workflows. - Provides platform-agnostic rules for when and how to delegate tasks to subagents vs. handling directly or scheduling. - Includes a five-step framework: task decomposition, delegation, parallel scheduling, validation, and wrap-up, with detailed quality checklists. - Features an MIT-licensed, zero-dependency delegation decision script for automated task assignment recommendations. - Extensive documentation covers best practices, common pitfalls, delegation patterns and anti-patterns, as well as platform mapping guides. - Tailored for platforms such as Claude Code, Codex, Cursor, WorkBuddy, and Hermes with tool mappings and practical field notes.