babyGao/agent-pilot-skills

# Dispatching GPT via Codex MCP

## Overview

You dispatch a task to GPT running as a Codex agent. It has its own sandbox and blank context — it sees nothing of your session, so you construct exactly what it needs. The value is that it is a *different model*: second opinion, adversarial review, or a parallel executor.

**Core principle:** One `codex` call = one independent GPT agentic turn. Reuse the `threadId` to continue; review its diffs before trusting them.

**Prerequisite — set it up on first use.** This skill needs the tools `mcp__gpt-codex__codex` and `mcp__gpt-codex__codex-reply`. If they're absent, provision them now, then tell the user to restart Claude Code (MCP tools only load at startup):

1. **Codex CLI + auth** — run `codex login status`. If Codex isn't installed or isn't logged in, tell the user to run `npm i -g @openai/codex && codex login` (interactive browser auth — only they can complete it).
2. **Register the server yourself** (user-level):
   - macOS/Linux: `claude mcp add gpt-codex -s user -- codex mcp-server -c sandbox_mode=workspace-write -c approval_policy=never`
   - Windows (run from PowerShell, or prefix `MSYS_NO_PATHCONV=1` in Git Bash — else `/c` mangles to `C:/`): `claude mcp add gpt-codex -s user -- cmd /c codex mcp-server -c sandbox_mode=workspace-write -c approval_policy=never`

# hello-boss

Cold outreach breaks at both ends. You cannot **reach** people — you do not know which
industries could even use what you sell, or which legal entity actually operates a brand, or
which address a decision-maker reads. And you cannot **land** — mail bounces or lands in spam.
The middle part, writing well, is the easy part, and it is the only part most tools help with.

Four stages, three deliverables, then it stops.

```
1. Demand discovery -> 2. Data mining -> 3. Message design -> 4. Send
   industries worth       companies +       one cold email      batch it out
   selling to             products + emails
```

# 火盘商业地产房源检索

火盘房源库的对外检索入口，已上线的 MCP 服务，只有一个工具 `search_listings`。
把用户的原话整句传进去，服务端自行拆成检索条件，返回结构化房源。

---

# 一、输出格式

**回复由三部分按序组成：**

1. **一行汇总** —— 命中总数、展示条数、实际生效的条件（取自 `applied_conditions`）
2. **每套房源一张卡片** —— 结构照下面这份原样套用，只替换数据
3. **一句收尾** —— 只说返回值支持的事实：结果跨了哪些维度（`unspecified`）、

# Orchestrating Parallel Research

## Overview

You run a fleet of agents as the **coordinator (军师), not a worker**. Your leverage is deciding *what to split, how to isolate, what to hand each agent, and how to synthesize* — never doing the grunt work yourself.

**Core principle:** front-load the standardization (directories, isolation, dispatch contract) so N agents run at once without colliding — then observe and adjust. It is a standing engineering setup decided **up front**, NOT mid-flight patching. Don't fix agent-1's layout at minute 10 then agent-2's at minute 20 — you predicted the situations and standardized before the first launch.

## When to Use

```dot
digraph when {
  "≥2 independent tasks?" [shape=diamond];
  "Shared state / tight sequence?" [shape=diamond];

# Publishing to 博客园 (cnblogs)

## Overview

Reliably publish a Markdown article to 博客园, or cross-post one from another platform (51cto, etc.). The browser is used only to **log in and (optionally) scrape**; the actual publish goes through cnblogs' own **REST API**, which is far more robust than driving the editor with selenium `send_keys`.

**Core principle:** Re-host every image on cnblogs' own CDN, and publish via the API with `isMarkdown: true`. Hotlinked source images break (anti-leech) and the 2024-era `send_keys` tool rots as the editor DOM changes.

## When to Use

- Cross-posting an already-published article (e.g. on 51cto) to 博客园
- Publishing a local `.md` (with front matter) to 博客园
- Symptoms: images on the published post show **只有链接/裂图**, markdown renders as raw text, or a browser-automation publisher fails on missing selectors

# 手绘暖调科技叙事风格 · 设计说明

> 取样自 13 帧故事板缩略图（GPT 讲解课）。本文用文字完整说明这套视觉风格的构成，并重点拆解**每一种手绘元素的画法**，可作为插画 / 动画落地的执行规范。

---

## 一、一句话定调

**把冷冰冰的科技，画成一张有人情味的手绘讲义。**

这套风格服务于「AI / 科技」叙事。它刻意回避矢量的精确与商业插画的光鲜，转而用**松弛的马克笔线条 + 扁平纯色块 + 大量留白**，把抽象的「思考、搜索、生成、协作」翻译成温暖、可亲近、像有人在旁边给你边画边讲的画面。

核心气质关键词：**手绘 · 温暖 · 极简 · 叙事 · 纸感**。