Skills agent-tools
📦

agent-tools

Safe ⚙️ External commands🌐 Network access🔑 Env variables📁 Filesystem access

Run 150+ AI Models via CLI

Also available from: inference-sh-7,inferencesh,inference-sh-6,inf-sh,inference-sh-0

Access powerful AI models without GPU setup. Generate images, videos, and text through a simple CLI that connects to cloud-based AI services including FLUX, Veo, Claude, Gemini, and OpenRouter.

Supports: Claude Codex Code(CC)
🥉 75 Bronze
1

Download the skill ZIP

2

Upload in Claude

Go to Settings → Capabilities → Skills → Upload skill

3

Toggle on and start using

Test it

Using "agent-tools". Generate an image of a cat astronaut

Expected outcome:

Task started. Task ID: tk_abc123. Check status with: infsh task get tk_abc123

Using "agent-tools". Call Claude for a summary

Expected outcome:

Claude Response: [AI-generated summary of the requested topic]

Using "agent-tools". Search for AI news

Expected outcome:

Found 5 results: [list of relevant articles with titles and summaries]

Security Audit

Safe
v1 • 2/18/2026

All 194 static findings are false positives. The flagged patterns (shell backticks, hardcoded URLs, environment variable access, pipe-to-shell) appear in documentation markdown files as legitimate CLI usage examples. The skill is a wrapper for the inference.sh CLI service - network access and credential handling are required functionality for calling external AI services, not security vulnerabilities. The critical heuristic warning about 'code execution + network + credential access' is a false positive because this combination is expected and necessary for AI service integration.

5
Files scanned
554
Lines analyzed
4
findings
1
Total audits

Risk Factors

⚙️ External commands (5)
references/app-discovery.md:5-106 references/authentication.md:5-53 references/cli-reference.md:5-97 references/running-apps.md:5-141 SKILL.md:15-109
🌐 Network access (4)
SKILL.md:7-126 references/authentication.md:6-59 references/cli-reference.md:6-104 references/running-apps.md:129-148
🔑 Env variables (3)
references/authentication.md:30 references/authentication.md:53 references/cli-reference.md:72
📁 Filesystem access (2)
references/cli-reference.md:84 SKILL.md:31

Detected Patterns

Documentation Shell Examples (False Positive)Service Endpoint URLs (False Positive)API Key Environment Variable (False Positive)Pipe to Shell Installation (False Positive)Heuristic: Code Execution + Network + Credentials (False Positive)
Audited by: claude

Quality Score

45
Architecture
100
Maintainability
85
Content
50
Community
100
Security
91
Spec Compliance

What You Can Build

Developer Quick AI Access

Generate images, videos, or call LLMs without writing code or setting up local AI environments

CI/CD AI Integration

Automate AI tasks in build pipelines using environment variable authentication

AI-Powered Content Creation

Create marketing images, social media videos, and generate copy using multiple AI models

Try These Prompts

Simple Image Generation 
Use the agent-tools skill to generate an image with the prompt: 'a futuristic city at sunset' using the flux model
Video Creation 
Run the agent-tools skill to create a video of a drone flying over mountains using the google/veo-3-1-fast model
LLM Text Generation 
Use agent-tools to call the Claude Sonnet model via OpenRouter to explain quantum computing in simple terms
Web Research 
Use agent-tools to search for the latest developments in AI image generation using tavily/search-assistant

Best Practices

  • Use 'infsh app list --search' to discover available models before running
  • Generate sample inputs with 'infsh app sample' to understand required JSON structure
  • Set INFSH_API_KEY environment variable for CI/CD automation
  • Use --no-wait flag for long-running tasks to avoid blocking

Avoid

  • Do not share your API key in public repositories or logs
  • Avoid running many parallel tasks without checking rate limits
  • Do not assume all models support the same input format - check each app's requirements

Frequently Asked Questions

What is inference.sh?
Inference.sh is a platform that provides CLI access to 150+ AI models without requiring GPU setup or local infrastructure.
How do I authenticate?
Run 'infsh login' to authenticate via browser, or set the INFSH_API_KEY environment variable for scripted access.
Which AI models are available?
Models include FLUX, Veo, Claude, Gemini, Grok, Seedance, OmniHuman, Tavily, Exa, OpenRouter models, and many more.
Is there a free tier?
Check inference.sh pricing at https://inference.sh for current pricing and free credit offers.
Can I use this in production?
Yes, the CLI supports environment variable authentication suitable for CI/CD and production workflows.
How do I check task status?
Use 'infsh task get <task-id>' to check the status of any running or completed task.

Developer Details

License

MIT

Repository

https://github.com/inference-sh-3/skills/tree/main/skills/agent-tools/

Ref

main

File structure

📁 references/

📄 app-discovery.md

📄 authentication.md

📄 cli-reference.md

📄 running-apps.md

📄 SKILL.md