skill-cross-platform-publish · Cross-Platform Publishing
A one-draft-multi-post skill: one piece of content is adapted to each platform's conventions, then published to Xiaohongshu, Douyin, Bilibili, WeChat Official Accounts, Kuaishou, WeChat Channels, and Zhihu.
The value isn't copy-paste but per-platform re-adaptation: publish_dispatch.py first generates a dispatch plan (checking each platform's character/aspect/hashtag/type constraints), then the LLM rewrites the content per platform constraints and local culture, reframes video across landscape/portrait platforms, and finally delegates actual publishing to each platform's publisher skill. With a Profile, a persona-consistency check runs before publishing (below 80, deviations are reported but publishing isn't blocked).
Example invocation: "Post this camping guide to Xiaohongshu, Douyin, and Bilibili at once."
Full brief
Positioning
skill-cross-platform-publish is the "one draft, many posts" orchestrator: one piece of content, re-adapted per platform's conventions, routed to each platform's publisher skill. Adaptation is done by the LLM, constraint checks and routing by script, actual publishing by the platform publishers — a clean three-way split.
Core capabilities
- Dispatch planning:
scripts/publish_dispatch.py planvalidates the manifest against each platform's constraints (characters / aspect ratio / hashtags / content type) and outputs a publisher-routed dispatch plan with warnings. - Per-platform re-adaptation: trims titles/body, cuts hashtags, and reformats topics per the plan's constraint notes; localizes tone to each platform's culture; reframes video via video-reframe when crossing landscape/portrait platforms.
- Per-platform delegation: covers 7 platform publishers — Xiaohongshu (skill-xhs-publisher), Douyin, WeChat Official Accounts, Bilibili, Kuaishou, WeChat Channels, Zhihu; platforms without a deployed publisher are flagged to the user first.
- Pre-publish persona check: with a Profile, adapted content runs through skill-persona-check — below 80, the score, deviations, and fix suggestions are reported without blocking the publish.
- Post-publish trail: result notifications and UTM-tracked short links; successful publishes are auto-recorded into the content calendar + skill-publish-log by the publisher script — no manual re-entry.
Workflow
- Generate the dispatch plan (
planvalidates constraints + routes); clear every warning - Adapt content per platform (rewrite title/body/hashtags; reframe video across landscape/portrait first)
- Persona check before publishing (with Profile)
- Call each platform's publisher skill to execute
- Push result notifications, generate tracked short links; publish records auto-archive
Inputs & outputs
| Input | Notes |
|---|---|
| manifest | JSON: content (title / body / tags / media type) + platforms (target platform list) |
Output: per-platform publish results; records auto-written to the content calendar (_schedule.json) and skill-publish-log.
Boundaries with adjacent skills
| Skill | Lane |
|---|---|
| skill-cross-platform-publish (this) | Multi-platform orchestration: constraint validation + per-platform re-adaptation + publisher routing |
| Platform publishers (skill-xhs-publisher, etc.) | Actual publishing execution on a single platform |
| skill-publish-scheduler | Scheduled batch publishing by time |
| skill-publish-log | The record/query/statistics base for publish events |
Fit
- One piece of content going to multiple platforms in one pass
- Content that needs platform-culture-aware rewriting, not verbatim copying
- Synchronized distribution across a multi-account matrix
Before you start
- Each target platform's publisher skill must be deployed and working; account login states must be ready (each publisher has its own login flow).
- The source content (title, body, tags, media files) must be prepared before adaptation.
skill-cross-platform-publish is part of the Aiglade Skill library. Invoke it from the Aiglade chat box in plain language.