Introduction¶

Invoking Agnes AI’s three sets of capabilities typically implies consulting three separate documents: text models use Chat Completions, image generation uses images/generations, and video generation is an asynchronous task requiring manual polling for the video_id. The interface styles, authentication methods, and error codes vary, meaning the primary integration cost is spent on format alignment.

The Agnes ai (AI Commander) introduced below is an API invocation Skill built based on the official Agnes AI documentation. It consolidates these three sets of invocations into a single SKILL.md, describing endpoints, parameters, and troubleshooting methods in a unified format.

What is this¶

The Skill is named Agnes ai (AI Commander), with coordinates @user_02952367/agnes-ai-aizhihuiguan. It is published by user_02952367, sourced from the community, categorized under ai-agent, and is currently at version 1.0.1. It solves the problem of solidifying the multi-model, multi-endpoint invocation methods of Agnes AI into a single skill description. This allows Agents to initiate requests in a unified format when they need conversation, image generation, video generation, image understanding, or tool calling, rather than assembling parameters on the fly.

One important note: The Skill is tagged with requires_api_key=true, meaning you need to prepare an Agnes AI API Key yourself before invocation.

Core Functions¶

The Skill covers three models, with their capabilities and corresponding endpoints listed in the table below (sourced from the model table on the Skill page):

Key points:

1. agnes-2.5-flash uses the OpenAI-compatible Chat Completions interface; text conversation, code, Agent, tool calling, and image understanding all share this single endpoint.
2. agnes-image-2.1-flash supports text-to-image and image-to-image, emphasizing high information density and composition retention.
3. agnes-video-v2.0 uses /v1/videos. Text-to-video, image-to-video, and keyframe animation all follow an asynchronous task pattern: submit the task first, then query the result using the task_id / video_id.

Installation and Activation¶

First, regarding the verifiable details. The deployment information provided on the Skill page consists only of: Category (ai-agent), Tag (requires_api_key=true), and Version (1.0.1). The page does not provide specific installation commands, SKILL.md file content, or configuration fields.

Therefore, installation methods for Cursor, Codex CLI, and Claude Code could not be verified from accessible materials, and I will not speculate here. Following general conventions for SKILL.md Skills, activation usually requires configuring the API Key as an environment variable. The specific variable name and acquisition method should be based on the instructions provided within the Skill package.

Typical Usage¶

The Skill is designed to be triggered when users have the following needs: invoking Agnes AI for image generation, video generation, conversation, image understanding, tool calling, Thinking mode, or streaming output; or troubleshooting issues such as 400 / 401 / 429 / 503 errors and video_id queries.

It should be noted that the official page could not be accessed to verify the full text of SKILL.md during this writing. Therefore, the complete list of required/optional parameters, curl and Python examples, prompt structures, and error code lists cannot be recited item by item. The confirmed invocation skeleton is: text-based requests point to /v1/chat/completions, image requests point to /v1/images/generations, and video requests point to /v1/videos utilizing the asynchronous task submission and polling method. The specific Base URL value also did not appear in verifiable materials; please refer to the SKILL.md for details.

Applicable Scenarios and Notes¶

Suitable scenarios:

1. You are already using Agnes AI and want to unify the description of the three types of invocations under a single Skill to reduce document switching.
2. Agent workflows requiring text-to-image / image-to-image, or text-to-video / image-to-video / keyframe animation.
3. Requirements to use tool calling, Thinking mode, or streaming output within a conversational chain.

Limitations to note:

1. This Skill requires a self-prepared API Key, and the page is marked with requires_api_key=true; invocation is impossible without a Key.
2. The video interface uses asynchronous tasks; you need to poll after obtaining the video_id, and handle async failures and timeouts yourself.
3. The Skill page shows 0 installations, a rating of 100000, and 1192 downloads (as of this crawl). Since it is a community source, it is recommended to read through the SKILL.md within the package to verify parameters before use.
4. The official page was inaccessible at the time of writing. All parameter details mentioned above are not expanded upon; please refer to the documentation within the Skill package before integration.

Conclusion¶

The value of this Skill lies in converging the invocation methods for Agnes AI’s three models and three types of endpoints into a single skill description. This provides a unified source for parameter formats and troubleshooting paths, rather than having them scattered across three documents. If your Agent requires Agnes AI’s capabilities for conversation, image generation, or video generation, you can start verifying from the addresses below.

Official Address: https://api.skillhub.cn/user_02952367/agnes-ai-aizhihuiguan
SkillHub Page: https://www.skillhub.cn/skills/user_02952367/agnes-ai-aizhihuiguan