Turn raw expertise into content with a point of view.
A Codex Skill for angle discovery, evidence-led writing, audience-native adaptation, and verified X publishing.
Install · How it works · Example · 中文说明 · lcax.asia
Most people do not struggle because their English is imperfect. They struggle because the original post has no angle, no evidence, and no reason for a reader to care.
A translator preserves the words. A generic AI writer polishes the prose. A scheduler publishes on time.
PostAngle decides what is worth saying first.
It turns notes, interviews, articles, product updates, and operating experience into content built around a clear judgment—then rewrites it as a separate edition for each audience and safely carries it through publication.
PostAngle is informed by the editorial and delivery experience behind lcax.asia, including patterns observed across 300+ client engagements.
| Tool category | Optimizes for | What is usually missing |
|---|---|---|
| Translation | linguistic equivalence | a useful point of view |
| Generic AI writer | fluent prose | evidence and author authority |
| Social scheduler | timing and distribution | content judgment |
| PostAngle | angle → evidence → audience-native writing → verified publishing | — |
The open-source landscape already has strong publishing infrastructure such as Postiz and Mixpost, plus general writing assistants such as Wordflow. PostAngle is deliberately narrower: it is an editorial decision system for turning expertise into platform-native content, not another calendar or literal translator.
PostAngle uses the ANGLE method:
- Audience tension — name one reader and one costly problem.
- Non-obvious judgment — find the idea that changes how the reader sees the problem.
- Grounding — support it with data, cases, mechanisms, or direct experience.
- Line of argument — build one causal path instead of a pile of points.
- Execution — adapt to the audience, confirm, publish, and verify.
flowchart LR
A[Raw expertise] --> B[3 angle candidates]
B --> C[Evidence gate]
C --> D[Audience-native edition]
D --> E[Human confirmation]
E --> F[Publish and verify]
The method also applies three public delivery principles from lcax.asia:
- show completed work;
- let people see before they use;
- let results speak.
- Finds and scores multiple editorial angles before drafting.
- Converts expertise into a single defensible thesis.
- Builds hooks, argument structure, evidence, caveats, and calls to action.
- Produces genuinely different Chinese and English editions.
- Verifies current claims, people, affiliations, and X handles.
- Chooses a single post, thread, or Article based on the idea and account capability.
- Requires human confirmation before public actions.
- Verifies the final thread order, mentions, comments, and direct links.
- Blocks private paths, infrastructure details, credentials, and client data from publication.
Requirements:
- Codex with personal Skill support;
- browser-control capability only when you want PostAngle to publish;
- an existing authenticated X session for publication—PostAngle does not handle passwords or 2FA secrets.
git clone https://github.com/luohechentim/postangle.git
mkdir -p ~/.codex/skills
cp -R postangle/skill/postangle ~/.codex/skills/postangleRestart Codex if the new Skill does not appear immediately.
Invoke it explicitly:
$postangle Turn these interview notes into a sharp English X thread for global founders.
Or use natural language:
这篇文章观点不够强。先找三个内容角度，再改成适合中文和英文受众的两个版本，确认后发到 X。
PostAngle can stop after writing, or continue through browser-based publishing and delivery verification when the user asks.
PostAngle is an X-first editorial Skill. It is designed to make content decisions and safely operate an interactive publishing workflow.
It is not:
- an unattended posting bot;
- a multi-platform content calendar;
- a replacement for factual source material;
- a promise of reach, conversion, or business results;
- a system that bypasses login, OAuth, CAPTCHA, 2FA, or human approval.
If a central fact, attribution, or account identity cannot be verified, PostAngle stops before publication and returns the unresolved item to the user.
We added more AI agents and tools. Coordination became difficult. We changed to one planner and temporary workers. It reduced repeated work.
AI agents can improve efficiency, but coordination is important. A planner-worker model can help teams collaborate better.
More agents made my AI workflow worse.
The bottleneck was never model intelligence. It was coordination: no shared memory, no clear owner, and the same failed experiments repeated across tools.
The fix was organizational, not technical—one control plane, temporary workers, and a durable experience log.
The difference is not vocabulary. It is the angle, causal claim, and proof structure.
See examples/before-after.md for the full bilingual breakdown.
PostAngle does not request passwords, cookies, API keys, or verification codes. It uses an existing authenticated browser session and pauses for the user at login, OAuth, CAPTCHA, 2FA, and final publication confirmation.
A filled composer is not treated as success. Delivery is complete only after the live post and links are verified.
postangle/
├── README.md
├── LICENSE
├── assets/
├── examples/
├── launch/
└── skill/postangle/
├── SKILL.md
├── agents/openai.yaml
├── assets/
└── references/editorial-method.md
很多人写不好自媒体，不是因为中文或英文不够流利，而是因为：
- 没有明确写给谁；
- 没有值得传播的判断；
- 没有证据支撑观点；
- 只是把材料整理得更顺；
- 把自动发布误当成内容能力。
PostAngle 先判断“什么值得说”，再解决“怎么说、对谁说、在哪里发”。
它会先从原始素材中找出三个不同角度，用受众痛点、观点新意、证据强度和作者权威性进行筛选，再分别写成适合中文与英文受众的独立版本。英文不是中文逐句翻译，中文也不是英文模板的本地化。
这套方法来自 lcax.asia 在 300+ 客户项目中的内容与商业 IP 实践。公开版只保留可复用的编辑判断，不包含客户隐私与内部资料。
$postangle 先找角度，再把这份素材写成有观点、有证据、适合传播的中英文 X 内容。
Built from practice, not prompt folklore — lcax.asia
If you believe content should start with a point of view—not a translation prompt—star the repository and share a before/after example in Discussions.