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 flaqai/backlink_skills --skill wechat-writergit clone --depth 1 https://github.com/flaqai/backlink_skillsWrote 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/flaqai/backlink_skills/wechat-writer)<a href="https://agentmods.dev/skills/flaqai/backlink_skills/wechat-writer"><img src="https://agentmods.dev/badge/skills/flaqai/backlink_skills/wechat-writer/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/flaqai/backlink_skills/wechat-writer"><img src="https://agentmods.dev/badge/skills/flaqai/backlink_skills/wechat-writer.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00142 | $0.04230 |
| Opus 5 | $0.00071 | $0.02115 |
| Sonnet 5 | $0.00028 | $0.00846 |
| Haiku 4.5 | $0.00014 | $0.00423 |
Grade A, and why
wechat-writer 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.
Copies of this mod
1 near-identical copy found in the catalogue:
- wechat-writer — 100% identical, 0 lines differ
How it starts
The opening of the file, as written. The whole thing — 342 lines — stays where its author put it; the contents beside it link to each section on GitHub.
WeChat Writer
目标
把一个主题、链接、素材包或旧稿，整理成一篇准确、有判断、对读者有用、符合账号声音、适合微信移动端阅读的公众号成稿。
本 Skill 复用 writer 已有的事实核查、SEO 审计、humanization、图片打包和可选 R2 能力，但采用公众号优先的写作逻辑：
- 先明确读者问题、核心判断和证据边界，再考虑标题吸引力。
- 把事实、推断、观点和作者亲历分开管理。
- 默认连续完成任务书、研究、提纲、初稿、改稿和审稿；用户说“交互模式”时才在选题或提纲处暂停。
- “写一篇”默认只生成审过的本地成稿。配图、公共图片 URL 和外部发布都是独立动作。
- 不把微信搜索优化等同于 Google SEO，不按关键词密度堆词。
默认使用简体中文；用户指定其他语言或账号语言时跟随用户。
场景路由
先判断用户真正需要什么：
| 用户意图 | 执行范围 |
|---|---|
| “今天写什么”“给几个选题” | 只做选题与评分，不写正文 |
| “写一篇公众号文章” | 完整执行到本地审过的 article.md |
| “完整制作一篇” | 成稿 + 图片方案/图片 + 微信交付包，不自动发布 |
| “检查/审一下这篇” | 只审稿；用户说“优化/改好”时直接改稿并复审 |
| “模仿这个风格” | 先建立临时风格指纹，再写作；不复制原句或虚构原作者经历 |
| “学习我的修改” | 对比修改，输出候选风格规则；只有重复出现或用户确认后才写入稳定档案 |
| “推到草稿箱/发布” | 先完成并封存成稿；只有另有可用发布工具且用户明确授权时才能执行外部发布 |
如果内容主要面向 Google 搜索，使用 ../SKILL.md。如果主要面向 LinkedIn Article、LinkedIn newsletter 或 LinkedIn 长文，使用 ../linkedin-writer/SKILL.md。如果主要面向 Medium 或第三方英文长文平台，使用 ../medium-writer/SKILL.md。
按需加载参考资料
必须完整读取与任务有关的参考文件：
- 新文章、改写、事实型内容：
references/wechat-brief-and-evidence.md - 选题、标题、开头、结构选择：
references/frameworks-and-hooks.md - 公众号语言、移动端阅读、图片和文件交付：
references/mobile-writing-and-packaging.md - 审稿、评分、修改与通过门槛：
references/wechat-review-rubric.md - 事实密集、产品比较或当前信息：
../references/fact-check-and-style.md - 完成事实与结构修改后的自然化：
../references/humanization.md - 图片文件、相对路径和可选 R2：
../references/output-packaging.md - 使用 R2 时：
../references/r2-image-upload.md和../references/r2-security.md
如果 profile/style-profile.md 存在，写作前读取它。不存在时不要把缺少风格档案当作错误；使用清晰、克制、具体的默认编辑声音。创建档案时以 profile/style-profile.example.md 为模板。
不可妥协的边界
- 不编造数据、研究、引语、案例、产品能力、价格、发布日期或“业内消息”。
- 不把模型记忆、搜索摘要或第三方范文标成已核实来源。
- 只有用户在当前任务明确提供的经历、观察、原话或材料，才能写成作者亲历。
- 无个人材料时可以写第一人称判断，但不能虚构“我的朋友”“我测试了”“那天我在现场”等事件。
- 第三方范文只能校准结构、节奏和抽象风格特征，不能提供可复用的观点、句子、人物、对话或经历。
- 标题不能为了打开率添加正文无法兑现的数字、身份、结果、恐惧或保证。
- Humanization 不是 AI 检测规避。不得承诺“检测不出”或故意加入错误、口癖和假故事。
- 写作、配图、排版、上传和发布是不同权限层级。完成正文不代表获得外部发布权限。
公众号文章任务卡
写作前创建或推断一张不超过 12 行的任务卡，保存到 wechat-brief.md：
- 账号/栏目：账号定位或本次栏目；未知时写“未提供”。
- 目标读者：谁会在什么场景点开。
- 真正问题：读者此刻想解决、判断或理解什么。
- 读后交付：读完能知道、相信、比较或执行什么。
- 核心判断：全文必须证明的一句话。
- 新增价值：相对常见文章多提供什么证据、角度、条件或方法。
- 文章类型：痛点解法 / 故事 / 清单 / 对比 / 热点解读 / 观点 / 复盘。
- 证据要求：需要核实的 3-6 个核心主张。
- 个人材料：可用 / 不可用；列明可用材料范围。
- 反方与边界：最强反对意见、适用条件和不适用场景。
- 目标长度：默认 1500-3000 个中文字符，按题材调整。
- CTA：读者下一步；没有商业目标时使用自然行动建议。
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 · 342 lines · 142 tokens per session scan A 5607f549047a
wechat-writer is a skill published in the GitHub repository flaqai/backlink_skills (692 stars, last pushed 16d ago), licensed MIT. It adds 142 tokens to every session and 4,230 once invoked, about $0.0007 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
gingiris-seo-geo-agent
SEO/GEO Agent Operations SOP v2 — Rebuilt from real-world operations (analook.com 0→launch, GSC incident, DataForSEO monitoring, gingiris.tools post-migration recovery). What's new in v2: • Step-by-step onboarding (Q1→Q6), multiple choice wherever possible — no big forms • Mandatory daily SEO log — first item is…
gr-backlinks
Systematic backlink building for indie founders. Covers 5 channel types ranked by GEO + SEO ROI: Wikipedia (entities), authoritative media (PR), industry reviews (G2/Capterra/Product Hunt), Reddit/Quora discussions, and HARO/Featured.com expert quotes. Designed for 0→1 sites where domain authority is the bottleneck.…
90-day-seo-sprint
Run the Distribb 90-Day SEO Sprint - a founder-built 13-week playbook for shipping pre-launch SEO, core pages, a content engine, and a backlink starter stack on the way to compounding organic traffic. Use when the user asks for an "SEO sprint", "90-day SEO plan", "SEO tracker", "SEO roadmap", "where do I even start…
distribb
Distribb is an SEO platform that handles keyword research, original data research, content publishing to WordPress/Webflow/Shopify, high-DR backlink exchange network, link building outreach playbooks, internal linking, social media repurposing and posting, Google Business Profile management (live reviews, public…
rankgoat
Run SEO for a website through RankGoat - plan content, generate and approve blog posts, and track backlinks, Domain Rating, sitemap health, and Search Console data. Use when the user asks about their site's SEO, content calendar, blog posts, backlinks, or rankings and they have a RankGoat account.
narrative-drift-monitor
Use when the user asks to "check if our surfaces have drifted from the canon", "watch for competitor repositioning", or "define when we should reposition"; produces a drift report — self-drift per flagship surface vs the narrative-registry canon over time (via wayback.py, change history Measured with as-of dates)…