start — active retort, community, active, ide skills, framework, javascript, orchestration, phoenixvc, typescript, windows, Claude Code

v1.0.0

关于此技能

Windows-first, polyglot AI-orchestration framework — agent teams, skills, quality gates, and governance for real projects

# 核心主题

phoenixvc phoenixvc
[0]
[0]
更新于: 3/25/2026

Killer-Skills Review

Decision support comes first. Repository text comes second.

Reference-Only Page Review Score: 1/11

This page remains useful for operators, but Killer-Skills treats it as reference material instead of a primary organic landing page.

Review Score
1/11
Quality Score
44
Canonical Locale
en
Detected Body Locale
en

Windows-first, polyglot AI-orchestration framework — agent teams, skills, quality gates, and governance for real projects

核心价值

Windows-first, polyglot AI-orchestration framework — agent teams, skills, quality gates, and governance for real projects

适用 Agent 类型

Suitable for operator workflows that need explicit guardrails before installation and execution.

赋予的主要能力 · start

! 使用限制与门槛

Why this page is reference-only

  • - Current locale does not satisfy the locale-governance contract.
  • - The page lacks a strong recommendation layer.
  • - The page lacks concrete use-case guidance.
  • - The page lacks explicit limitations or caution signals.
  • - The underlying skill quality score is below the review floor.

Source Boundary

The section below is supporting source material from the upstream repository. Use the Killer-Skills review above as the primary decision layer.

实验室 Demo

Browser Sandbox Environment

⚡️ Ready to unleash?

Experience this Agent in a zero-setup browser environment powered by WebContainers. No installation required.

Boot Container Sandbox

常见问题与安装步骤

以下问题与步骤与页面结构化数据保持一致,便于搜索引擎理解页面内容。

? FAQ

start 是什么?

Windows-first, polyglot AI-orchestration framework — agent teams, skills, quality gates, and governance for real projects

如何安装 start?

运行命令:npx killer-skills add phoenixvc/retort/start。支持 Cursor、Windsurf、VS Code、Claude Code 等 19+ IDE/Agent。

start 支持哪些 IDE 或 Agent?

该技能兼容 Cursor, Windsurf, VS Code, Trae, Claude Code, OpenClaw, Aider, Codex, OpenCode, Goose, Cline, Roo Code, Kiro, Augment Code, Continue, GitHub Copilot, Sourcegraph Cody, and Amazon Q Developer。可使用 Killer-Skills CLI 一条命令通用安装。

安装步骤

  1. 1. 打开终端

    在你的项目目录中打开终端或命令行。

  2. 2. 执行安装命令

    运行:npx killer-skills add phoenixvc/retort/start。CLI 会自动识别 IDE 或 AI Agent 并完成配置。

  3. 3. 开始使用技能

    start 已启用,可立即在当前项目中调用。

! 参考页模式

此页面仍可作为安装与查阅参考,但 Killer-Skills 不再把它视为主要可索引落地页。请优先阅读上方评审结论,再决定是否继续查看上游仓库说明。

Imported Repository Instructions

The section below is supporting source material from the upstream repository. Use the Killer-Skills review above as the primary decision layer.

Supporting Evidence

start

安装 start,这是一款面向AI agent workflows and automation的 AI Agent Skill。支持 Claude Code、Cursor、Windsurf,一键安装。

SKILL.md
Readonly
Imported Repository Instructions
The section below is supporting source material from the upstream repository. Use the Killer-Skills review above as the primary decision layer.
Supporting Evidence
<!-- GENERATED by Retort v3.1.0 — DO NOT EDIT --> <!-- Source: .agentkit/spec + .agentkit/overlays/retort --> <!-- Regenerate: pnpm --dir .agentkit retort:sync -->

start

New user entry point. Detects repository state, shows contextual status, and guides users to the right command or team for their goal. Read-only — suggests commands but does not execute them.

Usage

Invoke this skill when you need to perform the start operation.

Role

You are the Start Agent. Your job is to orient users — especially new ones — by detecting the current state of the repository and their session, then recommending the most relevant next steps. You are a router, not an executor. You suggest commands; you do not run them.

Behaviour

  1. Detect context (silent — do not print raw detection output)
  2. Show status summary (concise dashboard)
  3. Offer guided choices (interactive triage based on detected state)

Phase 1: Context Detection

Gather these signals silently:

  • AgentKit Forge initialised?.agentkit/ directory exists
  • Sync has been run?.claude/commands/orchestrate.md exists
  • Discovery completed?AGENT_TEAMS.md exists at repo root
  • Orchestrator has prior state?.claude/state/orchestrator.json exists
  • Backlog has items?AGENT_BACKLOG.md exists with content beyond header
  • Active tasks?.claude/state/tasks/ contains JSON files
  • Current git branchgit branch --show-current
  • Uncommitted changes?git status --porcelain
  • Orchestrator phase — read phase from .claude/state/orchestrator.json
  • Lock held?.claude/state/orchestrator.lock exists

Phase 2: Status Summary

Print a concise status table:

ItemStatus
AgentKit ForgeInitialised / Not initialised
SyncUp to date / Needs sync / Never run
DiscoveryComplete / Not run
OrchestratorPhase N (name) / No prior session
BacklogN items / Empty
Active tasksN tasks / None
Branchbranch-name
Working treeClean / N uncommitted changes

Phase 3: Guided Choices

Based on context, present ONE of these flows:

Flow A — Brand new (no discovery, no orchestrator state): Suggest: /discover (explore codebase), /healthcheck (verify health), /orchestrate (full lifecycle), /plan (fix a bug or plan a feature), /project-status (see what's here). Tip: start with /discover.

Flow B — Discovery done, no active work: Suggest: /orchestrate (start a task), /project-review (audit codebase), /backlog (see work items), /check (quality checks). Include a team routing table mapping goals to team commands.

Flow C — Mid-session (orchestrator has active state): Show current phase, task, and branch. Suggest: /orchestrate (continue), /orchestrate --status (check state), /orchestrate <new task> (start fresh), /project-status (review progress).

Flow D — Uncommitted work detected: Remind about uncommitted changes. Suggest git diff, commit, or stash. Note they can also proceed.

Decision Guidance — Dynamic Team Routing

If the user describes a task or asks which team to use, build the routing table dynamically from the repo's actual team configuration:

  1. Read AGENT_TEAMS.md (if it exists — created by /discover). Parse team names, focus areas, and scope patterns from the team assignment sections.
  2. If AGENT_TEAMS.md doesn't exist, fall back to reading .agentkit/spec/teams.yaml. Parse each team's id, name, description, scope, and handoff-chain fields.
  3. If neither exists, use the /team-* command files in .claude/commands/ — extract team names and descriptions from the YAML frontmatter description field.

From the discovered teams, build a routing table with three columns:

I want to...TeamCommand
(inferred from team description/scope)(team name)/team-<id>

Map the team's description and scope patterns to plain-language "I want to..." rows. For example:

  • A team with scope apps/api/**, services/** and description "API, services, core logic" → "Build or fix backend/API logic"
  • A team with scope src/components/**, src/pages/** and description "UI, components, PWA" → "Build or fix UI components"

Omit meta-teams (like forge) unless the user specifically asks about creating new teams.

For tasks that span multiple teams, recommend /orchestrate which handles cross-team delegation automatically.

State Management

This command is read-only. It reads state files for context detection but does not create, modify, or delete any files.

Rules

  1. Do NOT run any commands on behalf of the user. Only suggest them.
  2. Do NOT modify any files. You are read-only.
  3. Keep the output under 40 lines total (status + recommendations).
  4. Use plain language — no jargon without explanation.
  5. If $ARGUMENTS contains a task description, skip to the routing table and recommend the right approach immediately.
  6. Always end with: "Type any command to begin, or describe what you want to do."

Project Context

  • Repository: retort
  • Default branch: main
    • Tech stack: javascript, yaml, markdown

Conventions

  • Write minimal, focused changes
  • Maintain backwards compatibility
  • Include tests for behavioral changes
  • Never expose secrets or credentials
  • Follow the project's established patterns

相关技能

寻找 start 的替代方案 (Alternative) 或可搭配使用的同类 community Skill?探索以下相关开源技能。

查看全部

openclaw-release-maintainer

Logo of openclaw
openclaw

Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞

333.8k
0
AI

widget-generator

Logo of f
f

为prompts.chat的信息反馈系统生成可定制的插件小部件

149.6k
0
AI

flags

Logo of vercel
vercel

React 框架

138.4k
0
浏览器

pr-review

Logo of pytorch
pytorch

Python中具有强大GPU加速的张量和动态神经网络

98.6k
0
开发者工具