ai-social-media-content

Overview

What is ai-social-media-content?

ai-social-media-content is a social media content workflow built for the inference.sh CLI. It helps you generate AI-powered videos, reels, shorts, images, thumbnails, captions, and hashtags tailored to platforms like TikTok, Instagram, YouTube, and Twitter/X.

The skill uses the infsh command-line tool to call hosted AI models such as:

• Google Veo for TikTok- and Reels-style video generation
• FLUX, Seedance, Wan for creative visuals
• Kokoro TTS for AI voiceover and audio content
• Claude for copywriting (hooks, captions, descriptions, hashtag suggestions)

You work from your terminal, sending prompts and parameters to inference.sh apps so you can quickly test and produce social-ready assets without wiring up your own model infrastructure.

Who is this skill for?

ai-social-media-content is designed for:

• Content creators and influencers who want fast ideas and AI-generated assets for TikTok, Instagram Reels, and YouTube Shorts
• Social media managers managing multi-channel content calendars and looking to standardize formats and aspect ratios
• Brands and agencies that need repeatable, scriptable workflows for UGC-style content, campaigns, and A/B tests
• Developers and growth teams who prefer a CLI-first setup and want to integrate AI content generation into scripts or automation

If you are comfortable running Bash commands or already use inference.sh, this skill gives you a curated, platform-aware setup for social content generation.

What problems does ai-social-media-content solve?

This skill focuses on practical social media production tasks:

• Format confusion across platforms – references for recommended aspect ratios, durations, and resolutions for TikTok, Instagram Reels/feed, and YouTube Shorts/thumbnails
• Time-consuming asset creation – quickly generate short-form videos, thumbnails, and images from a prompt
• Inconsistent hooks and captions – use AI copywriting via inference.sh to draft platform-specific captions and hashtag sets
• Manual, non-repeatable workflows – move from ad-hoc usage of online tools to reproducible CLI commands you can version, schedule, and automate

It is a good fit when you want to explore and standardize AI-assisted social content production using inference.sh, especially around short vertical video.

When is it not a good fit?

ai-social-media-content might not be ideal if:

• You need a no-code GUI: this skill is built around the infsh CLI and Bash usage, not a drag-and-drop web editor.
• You require deep video editing, timelines, or compositing: it focuses on model-based generation, not full NLE-style editing.
• You cannot or do not want to use inference.sh: the commands in the workflow depend on the inference.sh environment.

If you only want basic caption suggestions without touching a terminal, a browser-based social media tool may be more appropriate.

---

How to Use

1. Installation and prerequisites

ai-social-media-content is distributed as a skill within the inferen-sh/skills repository and expects the inference.sh CLI to be installed.

Step 1 – Install inference.sh CLI

According to the skill, you must have the infsh CLI set up. Follow the official install instructions from the repository:

# View CLI install instructions from the skills repo
curl -s https://raw.githubusercontent.com/inference-sh/skills/refs/heads/main/cli-install.md | less

Install and configure infsh as documented there (including any required authentication or API keys).

Step 2 – Log in to inference.sh

Once installed, log in from your terminal:

infsh login

This connects your local CLI to inference.sh so you can run apps like google/veo-3-1-fast for video generation.

Step 3 – Add the ai-social-media-content skill

From an Agent or skills-enabled environment, add the skill from the shared repository:

npx skills add https://github.com/inferen-sh/skills --skill ai-social-media-content

After installation, open the skill’s SKILL.md and related guide files for detailed patterns and command examples.

2. First run: generate a TikTok-style video

The skill’s Quick Start demonstrates how to generate a vertical, TikTok-style video with Google Veo using infsh:

infsh app run google/veo-3-1-fast --input '{
  "prompt": "POV walking through a neon-lit Tokyo street at night, vertical format 9:16, cinematic"
}'

Key points:

• google/veo-3-1-fast is the app name invoked through inference.sh.
• --input accepts a JSON payload as a string. At minimum, you provide a prompt describing the desired shot.
• The prompt includes style and format hints (e.g., "vertical format 9:16, cinematic") to steer the output for short-form platforms.

You can tweak the prompt to match your niche, for example:

• Product showcase
• Day-in-the-life influencer POV
• Event recap or travel snippet

3. Use platform format guidance

The skill’s documentation lists typical platform specs such as:

• TikTok – 9:16 vertical, ~15–60s, 1080x1920
• Instagram Reels – 9:16 vertical, ~15–90s, 1080x1920
• Instagram Feed – 1:1 or 4:5, 1080x1080 or similar
• YouTube Shorts – 9:16 vertical, <60s, 1080x1920
• YouTube thumbnails – 16:9 landscape

Use these as prompts and parameter hints when you call models through infsh, for example:

• Mention 9:16 vertical for TikTok/Reels/Shorts
• Mention 16:9 thumbnail or "YouTube thumbnail" for still images

This helps the AI generate assets aligned with the platform you are targeting.

4. Extend to images, thumbnails, and copy

While the Quick Start focuses on a TikTok-style video, the skill is designed to support broader content workflows using inference.sh-hosted models like FLUX, Seedance, Wan, Kokoro TTS, and Claude.

Typical extensions include:

• Image and thumbnail generation – call an image-generation app via infsh app run ... with prompts that mention "YouTube thumbnail" or "Instagram post" details.
• Voiceover and audio – use a TTS app such as Kokoro TTS through infsh to generate narration that can be layered over your AI video.
• Captions, hooks, and hashtags – route a text prompt through a copywriting model (e.g., Claude via inference.sh) to generate:
  • Short hooks for the first 2 seconds of a Reel or Short
  • Platform-specific captions and descriptions
  • Hashtag sets tailored to TikTok, Instagram, or YouTube

These steps are driven by the same CLI pattern: select the app, pass structured input, and iterate on prompts.

5. Where to look in the repository

After installing the skill, use the file tree to understand the workflow in more depth:

• SKILL.md – primary overview of ai-social-media-content and its intended use
• guides/social/ai-social-media-content (if present in your checkout) – supporting guides, rules, or examples related to social workflows

Use these as a reference when building your own Bash scripts or integrating with your existing content tooling.

6. Adapting the workflow to your setup

The repository is meant as a starting point, not a turnkey app. For production use:

• Wrap infsh commands in Bash scripts that reflect your naming conventions and content calendar.
• Store your favorite prompt templates for different platforms and content types (e.g., product launch, tutorial, behind-the-scenes).
• Integrate with your file storage or CI so generated assets are uploaded or organized automatically.

You should adapt the guidance and examples to your own constraints (team workflow, review requirements, brand safety) rather than copying everything verbatim.

---

FAQ

How do I install ai-social-media-content?

1. Install and configure the inference.sh CLI by following the cli-install.md instructions in the inferen-sh/skills repository.
2. Log in with:

infsh login

3. Add the skill to your Agent environment:

npx skills add https://github.com/inference-sh/skills --skill ai-social-media-content

4. Open SKILL.md inside the skill directory for the latest usage notes.

Do I need the inference.sh CLI to use this skill?

Yes. ai-social-media-content is designed around the infsh CLI. The Quick Start and example commands assume that inference.sh is installed and that you can run infsh app run ... from your terminal.

Which social media platforms does ai-social-media-content support?

The skill is oriented around:

• TikTok short vertical video
• Instagram Reels and feed posts
• YouTube Shorts and video thumbnails
• Twitter/X posts and related content

The platform format guidance in the documentation helps you generate assets optimized for these destinations.

What types of content can I generate with ai-social-media-content?

Using inference.sh models referenced by the skill, you can generate:

• Short-form videos (e.g., TikTok, Reels, Shorts)
• Images and thumbnails for YouTube or Instagram
• Captions, hooks, and hashtags via AI copywriting
• Audio/voice with TTS models (for narration or voiceovers)

The exact capabilities depend on the apps you call via infsh app run (e.g., Veo for video, other models for images or text).

Is ai-social-media-content suitable for non-technical users?

The skill is terminal-first and assumes you can run Bash commands. Non-technical users can still benefit if a teammate or developer sets up scripts and simple command wrappers for them. If you prefer a full GUI with drag-and-drop editing, a dedicated social media design tool might be more comfortable.

Can I integrate ai-social-media-content into automated workflows?

Yes. Because everything runs through infsh and Bash, you can:

• Schedule generation jobs via cron or CI
• Chain commands to generate video, thumbnail, and captions in one script
• Commit your prompt and parameter presets to version control

The repository provides patterns; you decide how deeply to integrate them into your own automation.

Where can I find more examples and updates?

Open the ai-social-media-content area in the inferen-sh/skills repository:

• Browse SKILL.md for the canonical description and Quick Start
• Explore any additional guides under guides/social/ai-social-media-content if present

For new models, updated apps, or extended social workflows, watch the upstream inferen-sh/skills repository on GitHub.