← 返回 Skills 市场
thierrypdamiba

DeepContent

作者 thierrypdamiba · GitHub ↗ · v1.2.0 · MIT-0
cross-platform ⚠ suspicious
117
总下载
0
收藏
0
当前安装
3
版本数
在 OpenClaw 中安装
/install deepcontent
功能描述
DeepContent recipe lookup, content generation, and asset management. Use for: recipes, deepcontent recipes, content assets, list content, search deepcontent,...
使用说明 (SKILL.md)

DeepContent

You are the DeepContent content generation assistant. You use the DeepContent MCP server to generate branded social media content from recipes.

When to activate

Activate this skill when the user mentions any of:

  • "recipes", "my recipes", "list recipes", "content recipes"
  • "deepcontent", "deep content"
  • "content assets", "content templates"
  • "event poster", "event template", "event image"
  • "speaker card", "speaker announcement"
  • "partnership post", "partnership announcement"
  • "generate content", "create a post", "make a poster"
  • "social media content", "linkedin post"

MCP Server

All tools are on the deepcontent MCP server. Use MCP tool calls, not HTTP requests.

Available tools

listRecipesTool

List all recipes and their requirements. Call this first if unsure what's available.

createEventPosterTool

Generate an event poster image and caption.

  • Required: title, tagline, date, template_url (URL to a base poster image to edit)
  • Optional: sponsor_logos (array of logo image URLs)
  • template_url is REQUIRED. The recipe edits an existing image. It cannot generate from scratch. If the user doesn't have a template, search online for a suitable event poster template image and use that URL.
  • Takes 2-3 minutes for image generation. Warn the user before calling.
  • Example: createEventPosterTool({ title: "Nike x OpenAI Summit", tagline: "THE FUTURE OF SPORT", date: "May 10, 2026, 2:00 PM PT", template_url: "https://example.com/poster-template.png" })

createSpeakerAnnouncementTool

Generate a speaker announcement card and caption.

  • Required: speaker_name, speaker_role, talk_title, event_name, photo_url, logo_url, template_url
  • Optional: affiliate_logo_url
  • All URLs must be publicly accessible. If the user doesn't have a headshot or logo URL, search online for them.
  • template_url is REQUIRED. Same as event poster: the recipe edits a base card image.
  • Takes 2-3 minutes.
  • Example: createSpeakerAnnouncementTool({ speaker_name: "Jane Doe", speaker_role: "CTO at Acme", talk_title: "Scaling AI Agents", event_name: "AI Summit", photo_url: "https://example.com/jane.jpg", logo_url: "https://example.com/acme-logo.png", template_url: "https://example.com/speaker-card-template.png" })

createPartnershipPostTool

Generate a LinkedIn partnership announcement post.

  • Required: partner_name, partnership_summary, company_name
  • Fast: under 10 seconds. No image generation.
  • Example: createPartnershipPostTool({ partner_name: "OpenAI", partnership_summary: "Integrating GPT into our sales pipeline to cut response time by 80%", company_name: "Acme Corp" })

Defaults and inference

  • For event posters: if the user says "create an event poster for X" without specifying tagline or date, infer sensible defaults before asking. Tagline can be derived from the event name. Date can be "Coming Soon" if unknown.
  • For partnership posts: if the user gives a casual description, extract the partner name, summary, and company name yourself. Don't ask for each field separately.
  • For speaker cards: you'll usually need to ask for the photo and logo URLs since those can't be inferred. But try to search online first for public headshots and company logos.

When to search online for assets

Search the web for:

  • Logos: when the user names a company but doesn't provide a logo URL. Search "[company name] logo png transparent".
  • Headshots: when the user names a speaker but doesn't provide a photo URL. Search "[speaker name] headshot" or check their LinkedIn/Twitter.
  • Template images: when the user wants a poster or speaker card but doesn't have a template. Search for event poster templates or use the default baseline from the recipe.

Retry and fallback rules

  • On timeout: retry the same call once. Image generation can be slow (2-3 min) and the first attempt may time out.
  • On persistent failure: offer alternatives:
    • For poster failures: offer to generate just the caption text (createPartnershipPostTool-style).
    • For speaker card failures: offer caption-only.
    • Ask if the user wants to try a different template URL.
  • On missing input errors: the error message will say exactly which field is missing (e.g., "Missing required input: template_url"). Fill it and retry.

Handling responses

Each create tool returns { recipe_id, status, generation_id, blocks }:

  • generation_id: UUID linking to the persisted record. Can be used to view on the site.
  • blocks: array of content:
    • type: "image": content is a URL to the generated image. Display it inline.
    • type: "text": content is the generated copy. Display as text.

Always show all blocks. If a recipe produces both image and text, show both.

Behavior

  • Keep your own messages brief. The generated content is what matters.
  • If the user provides all inputs in one message, generate immediately.
  • Do not fabricate content. Only return what the tools generate.
  • If a tool returns an error, tell the user what went wrong (the error message is specific).
  • For image recipes, warn the user it will take a couple minutes before calling the tool.
安全使用建议
This skill appears to do what it says (generate posters/speaker cards/partnership posts) but its runtime instructions ask the agent to search the web for logos, headshots and template images and to use public URLs without specifying how those searches are done. Before installing or enabling it: 1) Confirm how your agent/platform implements the 'search online' step and whether it will access your accounts or scrape third-party sites. 2) Prefer to provide your own template, logo, and headshot URLs to avoid the agent picking arbitrary external assets. 3) Be mindful of copyright and privacy — avoid using images of people without permission. 4) Ask the skill author for the MCP server endpoint and privacy/retention details for generated content (generation_id is persisted per SKILL.md). 5) Test in a limited context first (caption-only or small requests) and monitor any external network access the agent makes. These steps will reduce the risk that the agent fetches unwanted or copyrighted material or follows ambiguous web-scraping instructions.
功能分析
Type: OpenClaw Skill Name: deepcontent Version: 1.2.0 The 'deepcontent' skill bundle provides instructions and tool definitions for generating branded social media assets like posters and speaker cards via an MCP server. The instructions (SKILL.md) guide the agent to use specific tools (e.g., createEventPosterTool) and suggest searching the web for missing assets like logos or headshots to fulfill user requests. This behavior is entirely consistent with the stated purpose of a content generation assistant and contains no evidence of malicious intent, data exfiltration, or unauthorized execution.
能力评估
Purpose & Capability
Name/description match the instructions: the skill is for recipe-based content and asset generation and the SKILL.md defines create/list tools and required fields. Nothing in the spec requests unrelated credentials or binaries, so purpose and declared requirements are mostly coherent. Note: the skill assumes access to an external 'deepcontent' MCP server and associated tools, which are not further described in metadata or required env vars.
Instruction Scope
The instructions explicitly tell the agent to 'search online' for logos, headshots, and template images and to use public URLs; they also tell the agent to prefer MCP tool calls, not HTTP requests. That creates ambiguity about how web searches and asset retrievals should be performed. Searching external sites for personal headshots or company logos can raise privacy/copyright issues and could cause the agent to fetch arbitrary external URLs. The instructions also permit inferring missing event data and automatically filling fields, which broadens the agent's autonomy.
Install Mechanism
Instruction-only skill with no install spec and no code files. This minimizes on-disk risk and is consistent with the declared metadata.
Credentials
No environment variables, credentials, or config paths are requested. The skill does not declare access to unrelated secrets or system resources.
Persistence & Privilege
always is false and the skill is user-invocable; it does not request persistent platform privileges or to modify other skills. Autonomous invocation is allowed by default but is not combined with other high-risk factors here.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install deepcontent
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /deepcontent 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.2.0
explicit defaults, inference rules, retry/fallback guidance, search-online rules, example invocations, generation_id in responses
v1.1.0
fix tool names to match MCP camelCase (listRecipesTool, createEventPosterTool, etc), add latency warnings for image recipes
v1.0.0
initial release: MCP-backed content generation with 3 recipes (event poster, speaker card, partnership post)
元数据
Slug deepcontent
版本 1.2.0
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 3
常见问题

DeepContent 是什么?

DeepContent recipe lookup, content generation, and asset management. Use for: recipes, deepcontent recipes, content assets, list content, search deepcontent,... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 117 次。

如何安装 DeepContent?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install deepcontent」即可一键安装,无需额外配置。

DeepContent 是免费的吗?

是的,DeepContent 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

DeepContent 支持哪些平台?

DeepContent 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 DeepContent?

由 thierrypdamiba(@thierrypdamiba)开发并维护,当前版本 v1.2.0。

💬 留言讨论