Contribution offer: runnable helper scripts + prompt templates for image/video skills

[oh-summy]

Hi! 👋

I'm oh-summy, I've built a companion skill library that adds runnable helper scripts on top of the existing Agnes AI documentation skill.

What's in it

Your current agnes-ai-models skill is great as a coding reference for agents (model catalog, API patterns, troubleshooting). My library extends it with executable tooling that agents can call directly:

Image understanding — understand_image.sh calls agnes-2.0-flash
Analyze an image (local file or URL) and return structured text or JSON.

Image generation — generate_image.sh calls agnes-image-2.1-flash
Generate or edit images, save locally or return a URL.

Video understanding — understand_video.sh calls agnes-2.0-flash
Sample frames from a video and analyze them (requires ffmpeg).

Video generation — generate_video.sh calls agnes-video-v2.0
Text/image to video with automatic async polling and download.

Plus prompt templates and structured output schemas for each modality.

Why it's compatible

• Uses only bash, curl, jq — no heavy runtime
• Same base URL, auth header, env var as your skill
• Works with Claude Code, Kimi Code, OpenCode, Codex, Cursor, Gemini CLI

Could I contribute these upstream?

I'd be happy to merge the helper scripts + templates into your repo in whatever form fits:

• New files under agnes-ai-models/scripts/
• A new references/prompt-templates.md
• Or a standalone section in the existing skill

Just let me know what structure you'd prefer and I'll send a PR. If it's not a fit right now, no worries — I'll keep them in my fork.

Thanks for building Agnes AI! 🙏

[agnesai-admin]

Thanks for the thoughtful contribution offer. The runnable helpers and templates could be a valuable complement to the current documentation-focused skill.

Please submit a pull request rather than merging code directly. The preferred structure is:

• Place executable helpers in agnes-ai-models/scripts/.
• Add prompt templates and structured-output guidance in agnes-ai-models/references/prompt-templates.md.

Please make every helper use AGNES_API_KEY and support AGNES_BASE_URL. The routing guidance is:

• International service primary: https://apihub.agnes-ai.com/v1
• International service alternate: https://apihub.agnes-ai.cn/v1 when the primary route has a network, DNS, TLS, or connection-timeout failure
• China service: https://api.agnes-ai.cn/v1

Please also document dependencies, provide a minimal example and error handling for each script, avoid logging secrets, and ensure any download action is explicitly initiated by the user. Include the contribution's source and license information so we can confirm compatibility with this repository.

We look forward to reviewing the PR.

[oh-summy]

Thanks for the detailed guidance — I'll follow the structure exactly.

A quick note on provenance for transparency:

These scripts originated in my personal skill repository
(https://github.com/oh-summy/skills), which is a collection of
AI-agent skills I maintain for my own use. The Agnes helpers are my
own original work, written from scratch based on the public Agnes AI
documentation and API spec. That repository is MIT-licensed and will
remain so; it will also grow to include other, unrelated skills over
time.

I'm contributing these scripts to AgnesAI-Labs/skills as an
unconditional, irrevocable MIT-licensed contribution. Once merged, the
code belongs to this repository — you are free to modify, relicense,
redistribute, or remove it at any time with no obligation or
attribution requirement back to me or my personal repository. I will
never assert any additional rights, claims, or restrictions over this
contribution beyond what the MIT License already grants.

I'll open the PR shortly with the requested structure. Looking forward
to your review.