Writing & Editing

用于根据一段中文小说简介，规划、起稿、续写、改写中文网文。重点覆盖选材、构思、分卷、章纲、开头、节奏、章末、模仿检索、去 AI 味。

Read the previous agent's state at the start of a session and write yours at the end. Use at the start and end of every session, after any milestone, decision, failure or environment discovery, and whenever the user says "continue", "pick up", "where are we" or "wrap up".

Branch, commit and open pull requests in this repo. Use whenever creating a branch, writing a commit message, pushing, opening or updating a PR, or when CI fails on a PR.

Write or edit documentation, READMEs, PR descriptions and research prose in this repo's style. Use whenever producing prose meant for readers.

Build an animation from scratch, making the decisions in the order that determines whether it feels right — should it animate at all, what purpose, which tool, which properties, which curve and duration, how it interrupts, how it exits. Writes the implementation. Use when asked to animate something, add motion, make a component feel alive, or build a transition. For critiquing existing motion use review-animations; for auditing a whole codebase use improve-animations.

Copy and text skill for antislop. Use when writing or editing prose: headlines, tone, CTAs, and anti-AI-writing patterns. Load with the core.

Convert Markdown long-form content into a polished X Article publishing draft. Use when a user asks to convert, adapt, format, or optimize Markdown for X Articles, including producing a copy-ready rich-text HTML version and an edited Markdown version.

Builds, audits, and migrates Salesforce Flows with current platform elements (Collection Filter, Transform including sums, counts and joins, record and picklist choice sets, reactive screens, screen actions, before-save updates, custom errors, asynchronous and scheduled paths, platform event triggers and waits, agent actions) instead of loop-and-decision patterns. Use for any Flow work: writing or reviewing .flow-meta.xml metadata, Flow Builder build specs, screen, record-triggered, autolaunched or scheduled flows, migrating Process Builder or workflow rules, and troubleshooting flows.

Write one complete agent brief from the framework's template for a single piece of work (feature, fix, polish, translation, accessibility or security item), with context, scope, reserved migration slot, scratch database, ports, file ownership, the standing rules (all configured languages, WCAG 2.2 AA, light and dark, privacy, data residency, What's new, no AI attribution), the gate and the report format. Use when the user or the plan-wave skill needs a brief, or says "write a brief", "brief an agent", "delegate this".

>- Start-of-day and end-of-day routine for the shared record. Run `day start` before any work (pull the team's latest, then a briefing) and `day end` before stopping (publish today's writeups). Also: who synced recently, and moving the KG publisher role for travel.

Apply Clean Code, SOLID, and maintainability principles when writing, reviewing, or refactoring application code. Use when improving code quality, reducing duplication, simplifying complex logic, or restructuring code.

Behavioral guidelines to reduce common LLM coding mistakes. Use when writing, reviewing, or refactoring code to avoid overcomplication, make surgical changes, surface assumptions, and define verifiable success criteria.

网文工业化写作引擎（免费公开版）：小说大纲、章节生成、润色、P0-A/P0-B 分级合规质检、快速/深度两档质检（三档结论）、行文逻辑通顺度检测、叙事一致性强制检查、世界观物质/经济自洽检查、设定库跨章自洽、爽点节奏量化、AI味/文风漂移检测。适配番茄、QQ阅读网文平台，内置抽屉项目隔离。

Writes and edits text in simplified technical English, based on ASD-STE100. Use when writing docs, specs, plans, READMEs, comments or commit messages, or when the user asks for plain technical text.

Run the review station on the current branch. Calls the verifier, then the reviewer subagent, and writes a findings summary for the human. Use when a change is ready for review.

Vietnamese academic research, thesis planning, writing, revision and citation checks, adapted from Imbad0202/academic-research-skills. Use for làm luận, tiểu luận, luận văn, khóa luận, đề cương, tổng quan tài liệu and academic follow-ups. Prioritize the current assignment and confirmed project; use Harvard by default or APA when required. Exclude unrelated chat and repository maintenance.

Routes each task to a suitable model tier in Claude Code, Codex, or another agent that can start subagents on a chosen model (for example Gemini CLI or Antigravity CLI), following the user's routing preferences. Keeps small tasks local, delegates discovery, routine coding, and testing to cheaper subagents, and sends difficult decisions and risky reviews to a stronger one. Use when deciding whether to delegate, writing subagent handoffs, or setting routing preferences, or when asked to "save tokens", "use cheaper models", "fan out subagents", "stay inside my quota", or "efficient-frontier-routing help".

Content Research command - write a detailed content brief for a chosen opportunity. Use only when the user invokes cr-brief or asks for the content-research "brief" action. Entry point for the content-research skill.

>- Mentor a user who builds a project to learn how it works. They write the core; you do all other work and keep it fun. Use when the user says "teach me", "guide me", "mentor me", "hints only", "no spoilers", "Socratic", or "help me learn X by building it", or when a project file says the project is for learning. Do not use for normal feature work. Do not use when the first request is "just fix it" or "write it for me".

>- Create a Clanker Ship ticket for work the person describes, following the team's ticket instructions. Use when the person asks to file, create, open or write up a ticket, bug or task. Fetches the team's live instructions with get_agent_instructions and creates the ticket with create_ticket.

>- Comprehensive guide and operational framework for engineering world-class, human-grade GitHub READMEs and technical repository documentation. Use whenever creating, auditing, or refactoring a README for open-source tools, self-hosted services, developer utilities, or client-server applications. Enforces clear documentation guidelines, the 30-second rule, dual-audience narrative balance, copy-pasteable quickstarts, and high-signal visual showcases.

Use when the desk needs a side (LONG, SHORT, or ABSTAIN) and an invalidation level for a market, on a CEX (Bybit V5 linear) or DEX (Hyperliquid perps), from closed bars only. Writes the vote and a failure note; never sizes or sends orders.

Use when evaluating a skill OR plugin's lifecycle across agents: detecting capability degradation after a model update, judging whether a new skill/plugin is effective and fits the user, or auditing whether to keep/rewrite/retire one. Works in WorkBuddy (skills), DeepSeek harness (plugins), and other hosts via adapters. Triggers: '评测这个 skill', '评测这个插件', '模型升级后 skill 还灵不灵', '新 skill 好不好用', '这个 skill 适合我吗', 'deepseek 插件好不好用', '跨 agent 评估能力', '做个 skill 并按评测流程走'.

Use when evaluating a skill OR plugin's lifecycle across agents: detecting capability degradation after a model update, judging whether a new skill/plugin is effective and fits the user, or auditing whether to keep/rewrite/retire one. Works in WorkBuddy (skills), DeepSeek harness (plugins), and other hosts via adapters. Triggers: '评测这个 skill', '评测这个插件', '模型升级后 skill 还灵不灵', '新 skill 好不好用', '这个 skill 适合我吗', 'deepseek 插件好不好用', '跨 agent 评估能力', '做个 skill 并按评测流程走'.