Skip to content

aiworkskills/wechat-article-skills / skills/aws-wechat-article-main

aws-wechat-article-main

Verified

公众号运营|微信公众号|公众号一条龙|公众号全流程|自媒体运营|wechat automation|content pipeline|AIGC workflow — 公众号一条龙运营总控入口,选题→写稿→审稿→排版→配图→发布串联 8 个子 skill,单条指令完成整篇图文从 0 到上架。面向公众号小编、自媒体、品牌内容。触发词分层:**一条龙流程**「一条龙」「完整流程」「从头做」「从 0 到发布」;**新做新发**「帮我写篇公众号文章」「做一篇公众号文章」「我想发一篇」「帮我发一篇」「再来一篇」;**选题起点**「今天写什么好」「有什么好写的」「找个话题」「爆款选题」「热点选题」「起个爆款标题」;**策划起点**「内容日历」「系列策划」「专栏规划」「连载」;**流程恢复**「接着上次那篇」「继续昨天的」「继续上次的」「接着之前的进度」;**显式模型新写**「用 GPT 写一篇」「用 DeepSeek 写一篇」「把提纲写成文章」。子 skill(topics/writing/review/formatting/images/publish/sticker/assets)单独触发仅限对**已有产物**的修改场景(如"改标题""润色这段""排版""审稿""加封面""发布");新做/策划/多环节串联一律走本入口。

references/config.example.yaml

View on GitHub
# Account & workflow config (template)
# Copy to .aws-article/config.yaml; secrets go in aws.env at repo root (see env.example.yaml in this directory)
# Per-article metadata & publish_completed go in each article.yaml; field docs: skills/aws-wechat-article-main/references/articlescreening-schema.md

# ── Account & positioning ──
article_category:        # Account niche / positioning (e.g. tech / education / lifestyle); affects overall style & topic selection
target_reader: ""            # Target reader persona (e.g. college students / beginners / dev managers); determines vocabulary depth & examples
tone:                    # Tone (casual / professional / formal / sharp / humorous); affects voice & sentence style
update_frequency:        # Publishing frequency (e.g. daily / weekly / biweekly); used for scheduling hints
default_author: ""           # Default author name; falls back to this when a single article has no author set

# ── Topics & titles ──
topic_direction: "不写XX"          # Topic direction or exclusions (e.g. "avoid XX" / "more case studies"); guides topic generation boundaries
title_style: 干货型          # Title style preset (e.g. practical / story / listicle); preference for title generation
title_max_length: 25         # Max title length (in Chinese characters)
summary_length: "80-128"     # Target summary word count range (for model prompts & self-check; WeChat digest max 128 chars)
forbidden_title_phrases: []  # Banned title words / phrases (list); prevents clickbait or sensitive terms

# ── Writing ──
writing_style: 口语化        # Writing style (conversational / formal / methodology / case-study / narrative); affects structure & expression
paragraph_preference: 短段为主 # Paragraph preference (short / long / mixed); affects layout & line breaks
heading_density: 每节必有小标题 # Subheading density (required per section / moderate / minimal); used for structure validation
target_word_count: ""        # Target word count (e.g. "1800-2500" or "2000"); used for writing & validation
closing_block: ""            # Global closing block (Markdown string); leave empty if using a preset
forbidden_words: []          # Banned words list for body text; used in review & self-check
original_attribution: ""     # Attribution style (e.g. "original / authorized reprint / references")

# ── Preset defaults ──
# Presets live in .aws-article/presets/{type}/; see each directory's README.md
# Format: must be YAML string lists. [] = no default; single item = use directly; multiple items = candidate pool —
#   the agent must narrow to a single-item list in this article's article.yaml before running write.py / format.py
#   (write validates structure/closing; format validates format_preset). Do not use scalar strings.
# Priority: article.yaml > custom_* > default_*
default_structure:                # e.g. single or multiple candidates
  - AI Skill 爆款教程结构
  - AI Skills 实操爆文结构
default_closing_block: []         # e.g. - brand
default_title_style:
  - 技术实操指南型
default_format_preset:                # Template names → presets/themes|formatting/ (亲和 / 资讯 / 书卷 / 杂志 …)
  - 资讯
default_format_scheme: []             # Color scheme of the chosen template; see `format.py --list-themes`
                                      # Empty = every scheme the template declares is available; the agent
                                      # picks one per article. Selecting a template always brings all of its
                                      # schemes — the website does not narrow this.
default_cover_image_style: []   # Cover preset names → cover-styles/ (built-in + .aws-article/presets/cover-styles/)
default_article_image_style: [] # Body image preset names → presets/image-styles/
default_sticker_style: []         # e.g. - storyboard; sticker workflow — agent picks one from the list

# ── Custom presets (override defaults) ──
# Same format as default_*: YAML string lists. When set, takes priority over default_*.
# article.yaml still takes highest priority over both.
custom_structure: []
custom_closing_block: []
custom_title_style: []
custom_format_preset: []
custom_format_scheme: []          # Overrides default_format_scheme. Hand-edit only — the website never
                                  # writes it; leave empty unless you want to bar some schemes.
custom_cover_image_style: []      # Overrides default_cover_image_style; same paths as default_cover_image_style
custom_article_image_style: []    # Overrides default_article_image_style; loads from presets/image-styles/
custom_sticker_style: []

# ── Images ──
# Cover presets: cover-styles/ (separate from body presets/image-styles/). Lists: default_* then custom_* then article.yaml.
image_source: generated       # generated | user; user = use uploaded images directly (skip image_create.py)
cover_aspect: "2.35:1"       # Cover image aspect ratio (e.g. 2.35:1 / 1.91:1 / 16:9); used for generation & crop reference
cover_style: 简约             # Broad preference alongside cover presets; minimal / illustration / photo / brand-color
image_density: 每节一图       # In-article image density (one per section / moderate / few / optional)
caption_style: 有图注         # Caption style (with captions / no captions / key images only)
multi_image_count: 6          # Default image count for multi-image scenarios (for generation or layout reference)
# Business assets (product intros, screenshots, brand images) go in .aws-article/products/{product-name}/

# ── Review ──
review_required:              # Required review checks (checked items appear in self-check / review report)
  - 敏感词                    # Example: check for sensitive words
  - 错别字                    # Example: typos / common misspellings
  - 链接                      # Example: external link format / availability
  - 原创标注                  # Example: whether original / reprint attribution is present
custom_sensitive_words: []    # Custom sensitive words list (overrides the general list)
review_output_format: 分块详细 # Review output format (brief / detailed-by-section); affects report presentation


# ── Embeds ──
# Articles can reference {embed:profile:name}, {embed:miniprogram:name}, {embed:miniprogram_card:name}, {embed:link:name}
# Per-article article.yaml should only set embeds.related_articles (e.g. manual); deep-merged with global related_articles.
# Do not duplicate profile / miniprogram config in article.yaml.
embeds:
  profiles:                   # WeChat official account cards (for {embed:profile:name})
    - nickname: ""            # Matches the placeholder; also used as data-nickname
      alias: ""               # Only for gh_ legacy format; new format uses id/headimg instead
      id: ""                  # data-id (when both id & headimg are set, outputs mp-common-profile)
      profile_id: ""          # Same as id; use either one
      headimg: ""             # data-headimg (official account avatar URL)
      signature: ""           # data-signature
      service_type: "2"       # data-service_type
  miniprograms:               # {embed:miniprogram:…} text links
    - title: ""               # Matches the placeholder; also used as link text when applink is set
      appid: ""               # Mini Program AppID
      path: "pages/index/index" # Launch path
      applink: ""             # When set, generates a weapp text link
      image: ""               # Thumbnail for legacy mp-miniprogram card (when no applink)
  miniprogram_cards:          # {embed:miniprogram_card:…} mp-common-miniprogram cards (separate from the above)
    - title: ""               # Matches the placeholder
      appid: ""
      path: "pages/index/index"
      applink: ""
      card_title: ""          # Card title (defaults to title)
      avatar: ""              # Mini Program avatar URL
      imageurl: ""            # Card cover image URL
      imageurlback: ""        # Optional: back-side image URL
  related_articles:           # Past / related article links (for {embed:link:name}; actual list usually merged from article.yaml's manual)
    manual:
       - name: "第一篇的占位名"
         url: "https://..."
       - name: "第二篇的占位名"
         url: "https://..."              # Manually pinned article list [{name, url}]; used by {embed:link:name}

# ── Paths ──
drafts_root: drafts/          # Local drafts root directory (one subdirectory per article)
published_root: posts/published/ # Local published archive directory (for backup / sync; does not affect live content)
series_root: series/          # Series / topic root directory (holds series config & assets)


# ── Publishing & comments (WeChat credentials in aws.env under WECHAT_N_*; this section only selects method & slot) ──
ENABLE_WECHAT:                # Whether WeChat account is configured; true / false
publish_method: draft         # draft | published | none; none = no WeChat config, publish.py full skips API calls
wechat_publish_slot: 1        # Multi-account: which WeChat slot to use (1..wechat_accounts); single account can omit or set 1
need_open_comment: 1          # 1 = enable comments on publish; 0 = disable (subject to WeChat backend / API capability)
only_fans_can_comment: 0      # 1 = only followers can comment; 0 = everyone can comment (requires comments enabled)

# Whether publishing is complete
publish_completed: false      # true / false; whether publishing is done
# Assets and presets are stored under .aws-article/; see .aws-article/README.md

# ── Writing model (third-party LLM) ──
# Supports any OpenAI-compatible endpoint (official, proxy, or self-hosted)
# Writing spec: see .aws-article/writing-spec.md (copy from writing-spec.example.md in this directory)
writing_model:
  provider: ""                    # Optional; auto-detected from base_url / model if empty (openai | volcengine | qwen | gemini)
  base_url: ""
  model: ""                  # e.g. qwen-plus / qwen-turbo / qwen-max
  temperature: 0.7
  max_tokens: 4000

# ── Image generation model ──
# base_url must be the FULL endpoint path (script detects mode from path suffix):
#   - https://xxx.com/v1/images/generations   (DALL-E / gpt-image)
#   - https://xxx.com/v1/chat/completions     (Gemini etc. via proxy)
#   - Gemini native / Qwen native / Volcengine: use their full endpoint URL
image_model:
  provider: ""                    # Optional; auto-detected from base_url path if empty (openai | volcengine | qwen | gemini)
  base_url: ""                    # Full endpoint URL (see examples above)
  model: ""
  default_size: "1024x1024"
  default_quality: "standard"
  # How to request an aspect ratio on /v1/chat/completions endpoints.
  #   auto (default) — send Gemini's extra_body.imageConfig only for Gemini-family models
  #                    (gemini / nano-banana / imagen); other models keep the text hint and
  #                    get the target ratio by post-generation crop.
  #   imageconfig    — always send it (for a proxy serving Gemini under a custom model name)
  #   none           — never send it (for strict gateways that reject unknown body fields)
  # Ignored by /v1/images/generations, Volcengine and Qwen endpoints, which have their own size params.
  aspect_mode: auto

# ── WeChat accounts ──
wechat_accounts: 1                       # Number of WeChat accounts; default 1
wechat_api_base: "https://wechat.aiworkskills.cn"   # WeChat API base URL. Official api.weixin.qq.com enforces an
                                         # IP allowlist, which blocks calls from a laptop; this
                                         # reverse proxy avoids that. Set it to
                                         # https://api.weixin.qq.com to talk to WeChat directly.
wechat_1_name: ""                                                # WeChat account 1 display name
wechat_2_name: ""                                                   # WeChat account 2 display name
writing-skillsobra/superpowersUse when creating new skills, editing existing skills, or verifying skills work before deploymentStars: 322kUpdated Claude CodeCodexVerifiedInstall
xlsxanthropics/skillsUse this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .xltx, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like "the xlsx in my downloads") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
shipping-and-launchaddyosmani/agent-skillsPrepares production launches. Use when preparing to deploy to production, or when asking what needs to be in place before shipping. Use when you need a pre-launch checklist, when setting up monitoring, when planning a staged rollout, or when you need a rollback strategy.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
childhoodthedotmack/claude-memBabysit a shipped release after merge, publication, or deployment. Verify published packages, production behavior, installed plugins, and late CI or user reports until the observation window is clean; use babysit for PR readiness before release.Stars: 118kUpdated Claude CodeCodexVerifiedInstall