Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add qiuqiu19950918-hue/agent-flow --skill agent-flowgit clone --depth 1 https://github.com/qiuqiu19950918-hue/agent-flowWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/qiuqiu19950918-hue/agent-flow/agent-flow)<a href="https://agentmods.dev/skills/qiuqiu19950918-hue/agent-flow/agent-flow"><img src="https://agentmods.dev/badge/skills/qiuqiu19950918-hue/agent-flow/agent-flow/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/qiuqiu19950918-hue/agent-flow/agent-flow"><img src="https://agentmods.dev/badge/skills/qiuqiu19950918-hue/agent-flow/agent-flow.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00155 | $0.08874 |
| Opus 5 | $0.00077 | $0.04437 |
| Sonnet 5 | $0.00031 | $0.01775 |
| Haiku 4.5 | $0.00015 | $0.00887 |
Grade A, and why
agent-flow scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 10d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
This is a copy
83% identical to chat_with_agent — 365 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 339 lines — stays where its author put it; the contents beside it link to each section on GitHub.
agent-flow · 多智能体调度工作流
调用命令：
$agent-flow（Skill 用$前缀；/前缀为 Command 专用，不可用于 Skill）。 自动匹配：当任务描述与本description意图匹配时，ZCode 会自动询问是否加载本 Skill，无需手动输入$agent-flow。 配套文档：四个子 Agent 的完整配置见agents.md。
1. 核心思想
主 Agent（Orchestrator）不直接动手写代码 / 跑命令，而是调度子 Agent 完成「检索 → 执行 → 命令」流水线（执行层含 code-executor 静态编辑与 general-purpose 动态开发两个角色）。主 Agent 的职责是：
- 拆解任务 → 确定需要哪些子 Agent、按什么顺序。
- 分发指令 → 给每个子 Agent 一段自包含、边界清晰的 prompt。
- 广播透明 → 向用户实时汇报「正在调度哪个子 Agent、调用哪个模型」。
- 分级错误处理 → 子 Agent 失败时，按级别升级（重试 → 兜底转交内置 Agent）。
1.1 设计权归属与三级能力阶梯（v4·最高原则）
设计权归属：主 Agent 拥有所有影响正确性的设计决策——架构、算法、几何/物理推导、参数取值、物理极限判断、以及验收方案。子 Agent 是执行层，只做机械翻译（照抄蓝图为代码）与被明确授权的运行时执行（跑命令/跑断言/十类受限自愈），不做方案设计。这是"主=脑 / 子=手"哲学的严格执行。
实测依据（4 组对比实验）：主 Agent 把设计权下放给子 Agent（语义蓝图）时，产物质量完全取决于子 Agent 能力——强子 Agent 能兜住，弱子 Agent（flash）认知过载空响应、烧 220 万 token 无产出；而主 Agent 亲手做完所有设计后给精确蓝图，即便弱子 Agent 也"全一次过零自愈"。子 Agent 的命运由主 Agent 的蓝图精度决定，而非子 Agent 模型本身。故把"主做全部设计"从最佳实践提升为硬规则。
所有 correctness-critical 的实现工作，按能力阶梯逐级兜底（不是"放权阶梯"——任何一级都执行主 Agent 的精确蓝图，设计权始终在主）：
| 级 | 执行者 | 模型定位 | 何时用 |
|---|---|---|---|
| Tier1（首选） | code-executor（照抄蓝图为代码，无 shell）+ cmd-executor（跑命令/跑断言套件 + 十类受限自愈） |
最弱但最便宜 | 主 Agent 已把设计推到精确蓝图（old/new 或坐标级）时，一律走 Tier1 |
| Tier2（次级兜底） | general-purpose（更强的手，工具全集含 shell） |
中（>code/cmd-executor，<主 Agent） | 仅当 Tier1 在主 Agent 精确蓝图下仍无法收敛时，主 Agent 重新生成更细方案后再派。general-purpose 拿到的仍是精确蓝图，不是语义蓝图 |
| Tier3（主接管） | 主 Agent 亲自下场 | 最强 | Tier2 仍失败 / 趋势失控 / §4.3 L3 硬触发 |
路由翻转（相对 v3）：v3 把 general-purpose 列为"动态开发首选"，实测导致主 Agent 偷懒走"语义蓝图"放权设计。v4 把 general-purpose 降为 Tier2 兜底，Tier1（code-executor+cmd-executor）成为唯一首选，从机制上堵死放权。阶梯升级是换更强的手，不是放设计权。
1.5 三类子 Agent 差异化对待（重要原则）
不同类型的子 Agent，其结果的可验证性不同，主 Agent 应差异化对待：
| 子 Agent 类型 | 结果可验证性 | 主 Agent 处理策略 |
|---|---|---|
| 执行类（code-executor） | 高（改了哪些文件、能否编译，明确可验） | 信任委托：主 Agent 不亲自执行编写，仅验收子 Agent 的状态回报 |
| 命令类（cmd-executor） | 高（退出码 + stdout 即验证） | 信任委托：主 Agent 不亲自执行命令，仅验收子 Agent 的退出码回报 |
| 检索类（code-retriever） | 低（返回的是"理解和片段"，无法直接验证完备性/准确性） | 强制验收：走可信度判定 + 完备性校验（见 verification.md） |
What ships with it
6 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 10d ago First seen · 339 lines · 155 tokens per session scan A 0228a5e50daa
agent-flow is a skill published in the GitHub repository qiuqiu19950918-hue/agent-flow (2 stars, last pushed 29d ago), licensed MIT. It adds 155 tokens to every session and 8,874 once invoked, about $0.0008 per session on Opus 5. A static security scan graded it A with 0 findings. It is 83% identical to chat_with_agent, differing in 365 lines, and is treated as a copy.
Other skills, from other repositories
systematic-debugging
Use when encountering any bug, test failure, or unexpected behavior, before proposing fixes.
local-ai-agents
Build local-first AI agents that run entirely on a developer workstation with Microsoft Foundry Local and Qwen function-calling models. Covers Small Language Models (SLMs), the OpenAI-compatible local endpoint, sandboxed local tools, local RAG with Chroma, local MCP servers, hybrid cloud/local routing, and the…
next-cache-components-adoption
Turn on Cache Components in a Next.js app and resolve the blocking routes it surfaces. Use when the user wants to enable, adopt, or migrate to Cache Components, flip the cacheComponents flag, work through a flood of blocking-prerender / instant validation errors, run the cache-components-instant-false codemod, or…
insight-error-page
Write or audit an insight-kind error page for the Next.js dev overlay. Use when creating a new errors/ .mdx page, auditing an existing one, or checking that a page matches the framework fix cards. Covers page structure, title alignment, FixCard cards with Copy prompt button, code snippets, terminology verification…
next-cache-components-optimizer
Drive a Next.js route to instant navigation by setting up an agentic loop, under Cache Components / PPR, on initial load (hard navigation) and client-side navigation (soft navigation). Encode the goal as a failing @next/playwright instant() e2e and work it to green, one verified route at a time; the shipped test then…
next-partial-prefetching-adoption
Turn on Partial Prefetching in a Next.js app and work through the insights it surfaces. Use when the user wants to enable or adopt Partial Prefetching, flip the partialPrefetching flag, opt routes in with export const prefetch = 'partial', audit Link prefetch={true} behavior, preserve existing prefetched UI with…