multi-agent SKILL.md skillA

multi-agent from frizzlefur/flowkit is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.

Install it, pinned
mdr add frizzlefur/flowkit/multi-agent@git:20260911.6d0310a

Pins git:20260911.6d0310a in mdr.lock. Pin the fingerprint instead.

Audit grade
A, 17 of 17 checks passed
Kind
SKILL.md skill
Repository
frizzlefur/flowkit
Path
skills/multi-agent/SKILL.md
Versions recorded
6
Latest version
git:20260911.6d0310a, committed Sep 11, 2026
First version recorded
Aug 28, 2026
Size
23,886 bytes
Repository stars
39
License
none declared

Agent Teams 方案生成与执行引擎。通过 Agent(name) + SendMessage(to: name) 工具链并行分发多 agent 团队(subagent 后台运行); 在 tmux 中且 pane 可用时自动获得分屏可视化,pane 故障或无 tmux 时静默降级为无分屏并发(无需任何前置依赖)。 当用户说 /MultiAgent、"fan out subagents"、"fan out"、"sends a team"、"多agent"、"团队协作"、"并行处理"、 "teammate"、"创建agent团队"、"派团队"、"扇出"、"并行深挖"、"派几个agent分头调研"、"digs deep"、"每个都深挖"、 "别漏掉任何东西" 时使用—— 即使只说一句 "fan out subagents" 也应触发本 skill(只读任务走 Fast Path 直接分发,写入任务走完整方案确认)。 支持项目上下文感知、协作式方案生成、分片覆盖验收(nothing missed)、环境自适应。.

Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.

What the file says

# MultiAgent Skill

## When to Use

Trigger when user:
- 显式调用 `/MultiAgent <任务描述>`
- 请求创建 Agent Teams / spawn teammates
- 使用关键词: "多 agent", "团队协作", "并行处理", "teammate"
- 说官方 Tip 同款短语: "fan out subagents", "fan out", "sends a team", "digs deep", "每个都深挖", "别漏掉任何东西"
- 中文口语委托: "派团队", "扇出", "并行深挖", "分头调研"
- 任务需要多 Agent 并发执行

## Core Architecture

```mermaid
graph TB
    A[用户输入] --> R{"风险路由: 只读 or 写入"}
    R -->|"只读/低风险"| F["Fast Path: 分片分解 + 直接分发"]
    R -->|"写入/高风险"| S0[Step 0: 项目上下文感知]
    S0 --> S1[资源检测]
    S1 --> S2[任务分析 + 角色匹配]
    S2 --> S3[协作式方案生成]
    S3 --> S4[用户微调确认]
    F --> S5[执行]
    S4 --> S5
```

## 路由:Fast Path vs Full Path

分发前先做一次风险判定,决定走哪条通道。判断标准是**任务性质**(只读 vs 写入),不是触发词本身——同一句 "fan out subagents" 对调研任务是 Fast Path,对改代码任务是 Full Path。

判定必须有显式锚点(防跳过,同 tmux 检测判定行机制):分发前在回复中写出
`路由判定: 只读 → Fast Path`(或 `写入 → Full Path`)——未判定就分发属于流程违规。

| 通道 | 适用 | 流程 |
|------|------|------|
| Fast Path | 只读/低风险:调研、信息收集、代码审查、文档阅读、多源比对 | 轻量上下文 → 分片分解 → 一行方案预告(告知式)→ 直接分批分发 → 分片清单核对(named agent 收尾同样执行「完成即总结即收」纪律) |
| Full Path | 写入/高风险:写代码、改配置、批量文件操作、跨系统重构 | Step 0-5 完整流程(项目上下文 → 角色匹配 → 协作式方案 + 用户确认 → 执行) |

### Fast Path(一句话 fan-out 场景)
…

Read the whole file at its exact version.

How to install

The exact contents, by fingerprint
mdr add frizzlefur/flowkit/multi-agent@sha256:2a4a1d9b8cc416be

The line at the top of this page pins the label git:20260911.6d0310a, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.

Badge

mdr badge

[![mdr](https://markdownregistry.com/badge/art_2pnie5ep2lnr4cp3.svg)](https://markdownregistry.com/a/art_2pnie5ep2lnr4cp3)

4 badge image fetches in 30 days. What the badge shows.

Versions

The registry has recorded 6 versions of this file, between Aug 28, 2026 and Sep 11, 2026, about one every 3 days.

versioncommittedcommitsizeaudit
git:20260911.6d0310a latest2026-09-11 6d0310a 23,886 BA view · diff
git:20260908.0bf68302026-09-08 0bf6830 23,834 BA view · diff
git:20260907.a401ae82026-09-07 a401ae8 23,642 BA view · diff
git:20260831.e8485d92026-08-31 e8485d9 22,417 BA view · diff
git:20260828.7733cea2026-08-28 7733cea 21,450 BA view · diff
git:20260828.dcab50d2026-08-28 dcab50d 21,331 BA view

Audit of the latest version

A  17 of 17 checks passed. Fixed automatic checks with no AI involved, so the same checks give the same answer every run.
  • pass: Frontmatter block present
  • pass: Frontmatter declares a name
  • pass: Frontmatter declares a description
  • pass: Size between 200 bytes and 200 KB (23886 bytes)
  • pass: No zero-width or bidi control characters
  • pass: No instruction hidden inside an HTML comment
  • pass: No link to an exfiltration or paste host
  • pass: No credential-shaped string
  • pass: No instruction to send local credentials anywhere
  • pass: No text hidden with inline styles
  • pass: No prompt-injection phrasing
  • pass: No curl or wget piped into a shell
  • pass: No recursive delete of root, home or parent
  • pass: No instruction to read or print local credentials
  • pass: No base64 blob over 200 characters
  • pass: No link to a raw IP address
  • pass: No script tag

Source

GitHub

frizzlefur/flowkit · 39 stars · license none declared · pushed 2026-10-11 · branch main

API

GET https://markdownregistry.com/api/v1/artifacts/art_2pnie5ep2lnr4cp3
GET https://markdownregistry.com/api/v1/resolve?ref=frizzlefur/flowkit/multi-agent
GET https://markdownregistry.com/api/v1/blob/2a4a1d9b8cc416beaa150171e8ac2b2a12d7672adaeb5a174c4ab56e7368aff5

Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.

Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.

More from frizzlefur/flowkit

AGENTS.md AGENTS.md file
frizzlefur/flowkit · AGENTS.md
git:20260909.4826024 · audit A · 39 stars
CLAUDE.md CLAUDE.md file
frizzlefur/flowkit · CLAUDE.md
git:20260909.e227067 · audit A · 39 stars
flow-deep SKILL.md skill
frizzlefur/flowkit · evals/flow-deep/skill-snapshot/SKILL.md · 当用户说 /flow-deep、深度编排、全量管道、这个任务很重要、要仔细处理、确保万无一失时使用此技能。 完整管道:superpowers检查→Prompt优化→深度思考→规划→代码级细化→并发执行→验证,自动注入 TDD 和代码审查。…
git:20260909.28c9794 · audit A · 39 stars
auto-skill SKILL.md skill
frizzlefur/flowkit · skills/auto-skill/SKILL.md · CRITICAL PROTOCOL: 本技能是所有任务或其他技能的底层依赖。严禁在未读取 auto-skill/skill.md 的情况下启动任何开发任务。这不是建议,而是必须执行的初始化程序。任何跳过此步骤的行为都将被视为任务失败。任何任…
git:20260908.0bf6830 · audit A · 39 stars
flow-deep SKILL.md skill
frizzlefur/flowkit · skills/flow-deep/SKILL.md · 当用户说 /flow-deep、深度编排、全量管道、这个任务很重要、要仔细处理、确保万无一失时使用此技能。 完整管道:superpowers检查→Prompt优化→深度思考→规划→代码级细化→并发执行→验证,自动注入 TDD 和代码审查。…
git:20260926.546fab1 · audit A · 39 stars
flow SKILL.md skill
frizzlefur/flowkit · skills/flow/SKILL.md · 当用户说 /flow、复杂任务、多步骤任务、编排、先优化再规划再执行时使用此技能。 将复杂任务拆分为"Prompt优化→深度思考→规划→并发执行"管道,支持 --quick/--think/--mermaid/--discuss 等参数。…
git:20260908.0bf6830 · audit A · 39 stars
prompt SKILL.md skill
frizzlefur/flowkit · skills/prompt/SKILL.md · Prompt 评分与优化工具:基于乔哈里视窗四象限定位信息盲区(第四象限独有知识必须用"喂模式"喂给 AI,否则 2 分封顶)、3S 原则(Single/Specific/Short)约束表达,输出 1-10…
git:20260908.0bf6830 · audit A · 39 stars

Every file in frizzlefur/flowkit

Other files named multi-agent

multi-agent SKILL.md skill
opencrabs/opencrabs · src/docs/reference/templates/skills/multi-agent/SKILL.md · Orchestrate a task across multiple AI models/providers. Define a team of agents with different roles, providers, and…
git:20260618.3ad5c24 · audit A · 1,020 stars
multi-agent SKILL.md skill
shibing624/agentica · agentica/skills/bundled/multi-agent/SKILL.md · Run work across more than one agent — the task subagent, the delegate tool, background commands, and the user's other…
v2.0 · audit A · 359 stars
multi-agent SKILL.md skill
wangjs-jacky/jacky-skills · archived/multi-agent/SKILL.md · 多 Agent 协作,并行调用多个 AI 模型并合并最佳结果。适用于需要高质量答案、代码审查、风险对冲的场景。
git:20260405.06b2dd2 · audit A · 15 stars
multi-agent SKILL.md skill
milanglacier/pi-multi-agent-skill · skills/multi-agent/SKILL.md · Start other pi agents and delegate work to them. Use it when a task benefits from a multi-agent workflow, such as…
git:20261012.057ec55 · audit A · 8 stars
multi-agent SKILL.md skill
adolfousier/opencrabs · src/docs/reference/templates/skills/multi-agent/SKILL.md · Orchestrate a task across multiple AI models/providers. Define a team of agents with different roles, providers, and…
git:20260618.3ad5c24 · audit A · 1 stars

Browse by kind, by grade A, or by owner.