Verified Agent Skill record

md2wechat

Convert Markdown to WeChat Official Account HTML and manage related article tasks: preview, draft upload, cover and infographic generation, image-post creation, writer-style drafting, title suggestions, AI trace removal, and discovery of supported providers, themes, prompts, and layout modules. Also covers unpublished Zhihu, CSDN, Toutiao, or Tencent Cloud Developer Community drafts through a browser.

Tasks

Format a Markdown article into WeChat Official Account HTML using a compatible theme and layout modules chosen via CLI discovery
Create a WeChat article draft with cover via `convert --draft` after passing `inspect --json` readiness checks and only when the user explicitly requests it
Generate an article cover or infographic using `generate_cover` or `generate_infographic` when a bundled preset fits
Create an image-first or multi-image post using `create_image_post`
Get WeChat title candidates with `title suggest <article.md> --json`, which emits a host-Agent AI request and does not choose the final title
Draft in a creator style or remove AI traces using `write` or `humanize`

Inputs

A Markdown article file (e.g., `article.md`)
Optional Brand Profile at `~/.config/md2wechat/brand.md` (free-form Markdown, not parsed by the CLI)
A named theme, provider, prompt, or layout module selected via discovery
Cover image via `--cover` or `--cover-media-id` when draft creation is explicitly requested

Outputs

WeChat Official Account HTML from `preview` or `convert` (byte-identical final API HTML from a successful converter result)
Structured JSON diagnostics from `inspect --json`, including `data.readiness.targets` and `data.readiness.blockers`
Title-generation prompt request JSON from `title suggest --json` for a host Agent or external model
Generated cover or infographic images when image generation is requested
A WeChat article draft or image post only when upload/draft side effects are explicitly requested

Limitations and checks

AI mode (`--mode ai`) does not parse `:::module` syntax, so advanced layout cards will not render there; API mode is required for advanced layout modules
`title suggest --json` does not call a model, upload, create drafts, or write back to Markdown
`doctor --json` is local-only and does not perform live authentication, upload images, or create drafts
Only the `minimax` provider and its `image-01` model accept `--subject-reference`; the reference must be a publicly reachable http(s) portrait image URL (inline data URLs and local paths are rejected)
Draft creation requires a cover; a WeChat URL or `mmbiz.qpic.cn` URL cannot be assumed reusable as `thumb_media_id`
On `PREVIEW_ACTION_REQUIRED` or `PREVIEW_FAILED`, the invocation does not create or overwrite preview HTML, and any pre-existing explicit output path is stale
Before any convert, upload, or draft, read `inspect --json` `data.readiness.targets` and `data.readiness.blockers`; if the target is blocked, stop and report blockers instead of guessing from legacy booleans or `checks`
Verify theme compatibility via `themes list --json`: API mode uses only `type: api` and `selectable: true`; AI mode uses only `type: ai` and `selectable: true`; never use non-selectable collection descriptors as concrete themes