md2wechat: Publish Markdown to WeChat Official Accounts
Convert Markdown to WeChat MP HTML with themes, AI images, drafts, and multi-account support.
Evidence: SKILL.md explicitly requires explicit user confirmation before upload/draft/publish side effects, avoiding over-reach; clearly separates read-only, preview, conversion, and credential-required side effects; states that WeChat secrets are never printed, requires API key and validates before side effects. Configuration boundaries are detailed in README. Weaknesses: permission boundaries not fully verified (e.g., install script behavior), publisher identity unverified, impacting attribution completeness. Deductions: good permission and confirmation mechanisms, but unverified install script behavior and unknown publisher.
Evidence: SKILL.md instructions are self-consistent, provide clear failure-handling paths (e.g., run doctor, config show), and state stop-and-report when blocked. README shows extensive command usage. However, no tests or CI-specific tests for the skill are provided, and static review cannot verify actual execution. Score limited to 7 due to inability to execute key paths and thin error-handling depth. Deductions: no reproducible test environment, static review limitations.
Evidence: SKILL.md describes clear applicable scenarios (convert Markdown to WeChat HTML) and provides detailed usage boundaries (e.g., AI mode vs API mode, image generation scenario selection). Environment fit: Chinese-oriented, README is in Chinese, but core functionality (API mode) relies on external overseas service (md2wechat API), which may be affected by network restrictions. Deductions: core functionality depends on overseas service, boundary declaration not fully verified.
Evidence: README and SKILL.md are well-structured, provide installation instructions and multiple doc links; clear private license (modified BUSL-1.1), but versioning is incomplete (no changelog or clear version management, README only has badges). Maintenance responsibility visible via GitHub but not explicit. Deductions: missing changelog and clear versioning, maintenance responsibility unclear.
Evidence: SKILL.md has clear goal, provides clear commands, README shows approximate output format. But static review cannot verify output is directly usable without actual execution. Score limited to 7 due to static review limitations. Deductions: output not verified, value evidence limited.
Evidence: CI workflow and go.mod exist, but CI details not provided, cannot confirm coverage of skill's key paths. README has trend badges but no third-party evidence. Static review limits maximum to 4 because CI workflow exists but unverified. Deductions: CI content not provided, lacking cross-source verification.
- Core API mode depends on external service (md2wechat API), which may be affected by mainland China network restrictions; assess network reachability before use.
- Install script runs via postinstall; review script content before installing to avoid supply-chain risks.
- This skill uses a private license requiring authorization for commercial use; read LICENSE carefully.
- Publisher identity is unverified; assess risk yourself.
- Static review cannot verify actual execution; test key paths in your environment.
What does this skill do, and when should you use it?
This skill wraps the md2wechat CLI, providing an AI-agent-friendly workflow for creating and publishing WeChat Official Account articles. It converts Markdown to WeChat HTML, supports theme and layout selection, article inspection, preview, draft creation, title suggestions, image generation, and AI-trace removal. The skill emphasizes a discovery-first approach, using CLI JSON output to make agent decisions verifiable. It requires Node.js and an npm package, and the md2wechat binary must be on PATH.
The skill instructs agents to run md2wechat commands, including:
- inspect --json to check article metadata and publish readiness, reading data.readiness.targets/blockers.
- Discovery commands like themes list, layout list, providers list to fetch available themes, layout modules, and image providers.
- convert to turn Markdown into WeChat HTML, optionally with --upload/--draft for image upload or draft creation.
- create_image_post for image-based posts; generate_cover/generate_infographic for cover or infographic images.
- title suggest --json to generate title suggestion prompts; write/humanize for content generation or de-AI-ing.
- doctor --json and config show --json for local configuration checks.
- Content creators who want to format Markdown articles for WeChat and preview before publishing drafts.
- Teams managing multiple WeChat accounts, needing batch conversion and draft creation with per-account credentials.
- Agent developers integrating WeChat publishing into automation, using JSON discovery commands for verifiable decisions.
- Users needing article images or infographics, via built-in image generation or plan mode for host-agent execution.
- Authors wanting to remove AI writing traces and produce more natural-sounding content with `humanize`.
What are this skill's strengths and limitations?
- Rich JSON discovery commands make agent decisions verifiable and reduce guesswork.
- Supports API mode (deterministic HTML) and AI mode (lighter) — API mode suits batch publishing.
- 48 professional themes and 53 advanced layout modules (API mode) offer strong visual variety.
- Multi-account management, local read-only, does not leak secrets.
- Image plan mode integrates with host-agent image generation tools.
- Advanced layout modules only work in API mode, which requires an API key and possibly commercial licensing.
- API mode requires `MD2WECHAT_API_KEY`; Source Available license restricts commercial use without authorization.
- Some features (e.g., `write`, `humanize`) may depend on external models, but skill docs don't detail model configuration.
- The repo bundles multiple skills; this is the OpenClaw variant, so nuances may be missed.
How do you install this skill?
- Install Node.js (version meeting CLI requirements).
- Install CLI globally:
npm install -g @geekjourneyx/md2wechat. - Initialize config:
md2wechat config init. - The skill assumes md2wechat is on PATH.
- The skill file is at
platforms/openclaw/md2wechat/SKILL.md; place it per agent-skills standard.
How do you use this skill?
Trigger the skill in an agent, e.g., "Format article.md and create a draft." The skill first runs md2wechat inspect article.md --json to check readiness. For title suggestions, run md2wechat title suggest article.md --json and let the host model produce titles. Ensure WeChat credentials and, for API mode, MD2WECHAT_API_KEY are configured before publishing.