tonone 入门导览

面向初次使用者介绍 tonone 的核心概念、常用命令和入门路径,帮助快速了解其工作方式。具体命令应以当前安装版本为准。

仓库
jeremylongshore/tons-of-skills-marketplace

安装方式

先用 Prompt 核对来源,也可切换到安装命令或下载可用的安装包。

安装方式

用 AI 助手安装

复制提示词,让助手先核对固定版本和依赖。

查看 Prompt收起 Prompt
先检查当前 Agent 是否已安装 SkillPI Skill Finder。如果尚未安装,请按官方仓库的说明安装并启用:https://github.com/Findskills/skillpi。然后使用 Skill Finder 打开此 SkillPI 详情页并定位 "tonone-onboard":https://skillpi.com/zh/skills/deepskill-7b15ea929f06a92a718a3b3af961a30c61e638dfc8a3e47186a71ac31a95cbea。安装前核对上游来源和版本:https://github.com/jeremylongshore/tons-of-skills-marketplace/blob/6ea46b08dce892944db7c16f28d0b5308d402402/plugins/ai-agency/tonone/skills/tonone-onboard/SKILL.md。来源不一致时停止并说明差异。核对通过后,用 Skill Finder 安装目标 Skill,并报告安装路径和来源版本。

此命令从 GitHub 来源仓库获取指定技能。运行前请先查看源文件。

运行工具
npx skills add https://github.com/jeremylongshore/tons-of-skills-marketplace/tree/6ea46b08dce892944db7c16f28d0b5308d402402/plugins/ai-agency/tonone/skills/tonone-onboard --skill tonone-onboard
许可协议: —

来源与配置

查看 GitHub 源码

jeremylongshore/tons-of-skills-marketplace固定提交 6ea46b08dce892944db7c16f28d0b5308d402402

配置方式待核实

请查看原始来源中的适用工具、依赖和权限要求。

安装实测待补充

面向初次使用者介绍 tonone 的核心概念、常用命令和入门路径,帮助快速了解其工作方式。具体命令应以当前安装版本为准。

↓ 0 下载 ↗ 0 分享点击

文件浏览

在原站浏览文件 ↗

原始 SKILL.md

打开原文 ↗
---
name: tonone-onboard
description: 'First-run onboarding tour — guided walkthrough of tonone''s 23 agents, key skills, and worktree sessions. Two paths — expert (~90 sec) and newcomer (~8 min). Use when asked "how do I use tonone", "what can tonone do", "show me around", or "first steps".'
allowed-tools: AskUserQuestion
version: 0.8.0
author: tonone-ai <hello@tonone.ai>
license: MIT
---

# tonone-onboard

Cross-agent onboarding tour. Not tied to a single agent.

Always runs. Never checks the marker file — the skill replays the tour regardless
of prior runs. To re-show the SessionStart welcome banner, delete
`~/.config/tonone/onboarded`.

## Step 1: Tier Check

Ask via AskUserQuestion:

> Are you familiar with Claude Code agents?

Options:

- A) Yes — I know CC agents, just show me tonone's capabilities (~90 sec)
- B) No — walk me through the whole thing (~8 min)

---

## Step 2: Expert Path (A)

### What tonone is

23 specialists, 2 teams. Engineering (15 agents) + Product (8 agents). Each owns a
domain. You dispatch them. They don't fight over work — Apex routes automatically.

### Top 5 commands to bookmark

Output this block verbatim:

```
┌─────────────────────────────────────────────────────────────┐
│  /apex-takeover     hand any task to the full team          │
│  /atlas-onboard     generate project docs for day-1 devs   │
│  /forge-audit       infra cost check                        │
│  /relay-ship        deploy your stack                       │
└─────────────────────────────────────────────────────────────┘
```

### Mental model

**Worktree sessions:** Every session gets its own git branch automatically.
Parallel sessions never conflict. Clean sessions auto-remove their branch on close.

### Done

> Run `/apex-takeover` to start. Describe any task and Apex routes it.
>
> Replay this tour any time: `/tonone-onboard`

---

## Step 3: Newcomer Path (B)

### What Claude Code agents are

Claude Code can act as specialized agents — each configured with a persona, domain
knowledge, and a set of skills. Instead of one generalist AI, tonone gives you a
team of 23 specialists. You talk to them like colleagues. They coordinate through
Apex, the engineering lead.

### Meet the team

Output this block verbatim:

```
Engineering Team (15 agents)
─────────────────────────────────────────────────────────────
Apex    Engineering lead — routes tasks, coordinates the team
Atlas   Knowledge engineer — docs, ADRs, onboarding
Forge   Infrastructure — cloud, IaC, cost
Relay   DevOps — CI/CD, deployments, GitOps
Spine   Backend — APIs, system design, performance
Flux    Data — databases, migrations, pipelines
Warden  Security — IAM, secrets, threat modeling
Vigil   Observability — monitoring, alerting, SRE
Prism   Frontend/DX — UI, internal tools, portals
Cortex  ML/AI — LLM integration, evals, RAG
Touch   Mobile — iOS, Android, cross-platform
Volt    Embedded/IoT — firmware, edge, protocols
Lens    Analytics — dashboards, metrics, reporting
Proof   QA — test strategy, E2E, flaky triage
Pave    Platform — dev experience, golden paths

Product Team (8 agents)
─────────────────────────────────────────────────────────────
Helm    Head of Product — briefs, handoff to engineering
Echo    User research — interviews, personas, JTBD
Lumen   Product analytics — OKRs, funnels, A/B tests
Draft   UX design — flows, IA, wireframes
Form    Visual design — brand, tokens, design system
Crest   Strategy — roadmap, prioritization, competitive
Pitch   Marketing — positioning, messaging, GTM
Surge   Growth — acquisition, activation, retention
```

### How to invoke Apex

Run `/apex-takeover` and describe your task. Example:

> Run `/apex-takeover` and say "check our security posture."
> Apex reads the request, dispatches Warden, brings you the result.
> You never invoke Warden directly.

The right agent always gets the job. You don't need to know who to call.

### Worktree sessions

Every session gets an isolated git branch at `.claude/worktrees/`. Sessions editing
the same files don't conflict — they each work on their own branch. When a session
ends with no changes, the branch cleans itself up automatically.

You don't need to manage this. It happens automatically via the SessionStart and
Stop hooks.

### Skill routing

Skill routing tells Claude to use a specialized workflow instead of answering
directly when the request matches a known pattern. This is already configured in
this project's `CLAUDE.md` — see the `## Skill routing` section.

Examples already wired:

- "there's a bug" → `/investigate`
- "ship this" → `/ship`
- "review my diff" → `/review`
- "product idea" → `/office-hours`

You can add your own routing rules to `CLAUDE.md` at any time.

### Next steps — try these first

Output this block verbatim:

```
1. /apex-takeover     describe any task, Apex routes it
2. /atlas-onboard     generate onboarding docs for this project
```

### Done

> You're set. 23 agents, 100+ skills, isolated sessions.
>
> Replay this tour any time: `/tonone-onboard`
> Re-show the install banner: delete `~/.config/tonone/onboarded`

当前收录版本

0.8.0

核对来源版本 ↗

质量评测

满分 100。前四项来自 DeepSkill 收录的质量报告。

场景价值95/100
评分依据

高价值首次运行引导教程,直接解决用户上手核心痛点

触发精准95/100
评分依据

触发词精准完整,覆盖多种变体,边界清晰

结构规范95/100
评分依据

Phase/步骤清晰,有出口条件,含端到端示例

执行可靠95/100
评分依据

工具来源明确,参数正确,覆盖主要异常

证据可验 (前四项均值)95/100
计算方式

取前四项原始分值的算术平均,四舍五入取整。可通过收录报告核对各项评分与说明,运行记录需单独查证。

权限与使用前检查

根据当前主文档自动整理,命中内容可能是示例或防护说明;不是实际权限声明,也未覆盖附件、外部依赖或运行行为。

未从主文档识别到上述常见操作,不能据此判断无需权限或没有风险。

本站展示提示不会授予权限。涉及文件改动、密钥、外部发送或费用时,仍由你使用的工具按需确认。

安全评级与访问范围

risk · 静态证据核验,不代表运行安全保证。

使用前请由宿主核对并授权实际文件读写、外部服务和联网目的地;本站的核验不会授予任何运行权限。

来源与使用声明 SkillPi 的部分技能资料来自公开来源或创作者提交。第三方内容的权利归各自权利人所有。安装或运行前,请查看原站的许可、权限要求和使用风险。

来源与使用

请按详情页的作者、原始来源和许可证核对使用条件,运行前检查技能需要的权限。目录收录不等于已验证安装或安全性。