Agnes AI Skills
The Agnes AI Models skill helps an agent connect Agnes AI text, image and video models through an OpenAI-compatible gateway
- 68 ★ on the repository
- Last commit
Extended access
Why this level
- The skill works with a paid API: it needs a key from your account and usage is billed to your plan
- The API key lives in an environment variable and must not reach the repository
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/agnes-ai-skillsDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The skill works with a paid API: it needs a key from your account and usage is billed to your plan
- The API key lives in an environment variable and must not reach the repository
This is third-party code. Review the repository files before installing.
What it does
The AgnesAI-Labs/skills repository calls itself the official Agnes AI skill repository and holds one skill, agnes-ai-models, for Codex. The skill guides an agent through integrating the models over an OpenAI-compatible API: access setup, model choice, code examples and error diagnosis. The models in the README: agnes-2.5-flash for text and image understanding, agnes-image-2.5-flash for image generation up to 4K, and agnes-video-2.5 and agnes-video-2.5-flash for video at resolutions from 720P to 2K. It covers chat requests, streaming, image generation with editing through extra_body.image, video generation with polling by video_id and model_name, tool-calling workflows and notes on the Token Plan and RPM limits. The references folder holds a model catalog, a breakdown of common errors and a guide for other agents. For Codex the install goes through install-skill-from-github.py with --repo AgnesAI-Labs/skills and --path agnes-ai-models, after which Codex is restarted. For OpenClaw, Hermes, Manus and custom agents the skill files are read as instructions. You need an account on the Agnes platform and a key in the AGNES_API_KEY variable.
Who it is for. Developers who connect Agnes AI models to their app or agent and want Codex to write working call examples.
Good fit when
- You need a Python example for agnes-2.5-flash with a streamed response
- You need to generate or edit an image with agnes-image-2.5-flash
- A video has to be created with agnes-video-2.5 and polled for readiness by video_id
- An API error is unclear and you want the checklist from references/troubleshooting.md
- Your own agent supports OpenAI-compatible providers and needs the base URL, key and model name filled in
Not a fit when
- You need models from other vendors: the skill knows only Agnes AI
- You have no Agnes platform account and API key: calls will not pass without them
- You want to fix an authentication or quota problem by switching routes: the README forbids that
- You want a universal client for every OpenAI-compatible service
Example request
Use the Agnes AI Models skill to write a Python example for agnes-2.5-flash with streamingLimitations
- You need an account at platform.agnes-ai.com and an API key in AGNES_API_KEY, and the key must not be committed.
- Routes differ: international https://apihub.agnes-ai.com/v1, alternate https://apihub.agnes-ai.cn/v1 only for a network, DNS or TLS failure, and a separate one for China. Do not switch routes over authentication, quota or rate-limit errors.
- The repository is packaged as a Codex skill, and for other agents the files are read as instructions.
- The repository states no license.
- RPM limits and the Token Plan are described in the reference and depend on the account.
How to disable. Delete the agnes-ai-models folder from the Codex skills directory and the AGNES_API_KEY variable from the environment.
FAQ
Which models does the skill describe?
agnes-2.5-flash for text and image input, agnes-image-2.5-flash for image generation, and agnes-video-2.5 and agnes-video-2.5-flash for video. Each has its own endpoints and limit notes in references/model_catalog.md.
How do I install the skill in Codex?
Run python /path/to/install-skill-from-github.py --repo AgnesAI-Labs/skills --path agnes-ai-models and restart Codex. There is also a variant with --url pointing at the folder https://github.com/AgnesAI-Labs/skills/tree/main/agnes-ai-models.
What do I need to use it?
An Agnes platform account and an API key stored as AGNES_API_KEY. The international service base URL is https://apihub.agnes-ai.com/v1.
Does the skill work beyond Codex?
Yes, the skill files are read as instructions in OpenClaw, Hermes, Manus and other agents that support custom OpenAI-compatible providers: set the base URL, key and model name.
Which API routes does Agnes AI have?
International https://apihub.agnes-ai.com/v1 is primary, https://apihub.agnes-ai.cn/v1 is alternate only for a network, DNS or TLS failure, and China uses https://api.agnes-ai.cn/v1.
Discussion
No questions or reviews yet.
Related
SkillsA library of 30-plus instruction skills for coding agents: planning, building, testing, hallucination guards and shipping
The author's personal set: shared agent rules and dozens of skills for GitHub, the browser, macOS, messaging and releases, plus a sync script
Claude Code Skills and Agents
Claude Code Skills & Agents
A Claude Code plugin bundling skills and sub-agents: deep research, image generation, GitHub work, code review and model switching
Over twenty skills for coding agents: MCP and hooks, plan mode, a bug-fix protocol, prompting and a bash safety hook
agnix lints CLAUDE.md, AGENTS.md, SKILL.md, hooks and MCP configs with 702 rules, autofix and editor extensions for VS Code and JetBrains
19 skills for Claude Code, Codex and Antigravity CLI: Grok and Exa search, Context7 docs, Serena and task hand-offs between agents
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Agnes AI Skills (this card) | yes | 68 | |
| Agnes AI Generation Skill | no | 426 | a separate generation skill, while this is the vendor's official set with a model catalog, routes and error handling |
| Agnes Free Model Skills | no | 202 | a set about Agnes free models, while agnes-ai-models covers text, images, video and the Token Plan |
README in short
The README calls the repository the official store of Agnes AI skills, lists the agnes-ai-models skill, a model table, install methods, requirements with an API key and region routing. A quick start with an example prompt, the list of covered scenarios and official links follow.
README badge
Original SKILL.md
--- name: agnes-ai-models description: Agnes AI model integration skill for OpenAI-compatible text, image, video, and agent workflows. Use when Codex needs to configure Agnes AI API access, select or document Agnes models including agnes-2.5-flash, agnes-image-2.5-flash, agnes-video-2.5, and agnes-video-2.5-flash, write Python/Node/curl examples, debug common API errors, poll video results with video_id, adapt OpenAI-compatible clients to the Agnes API gateway, or prepare guidance for other agent software such as OpenClaw, Hermes, Manus, and custom agents. --- # Agnes AI Models Use this skill to help developers integrate Agnes AI models through the OpenAI-compatible API gateway. Users must register at `https://platform.agnes-ai.com/` and apply for an API key before making requests. Never invent, expose, or ask the user to paste secrets into public files. ## Core Defaults - API base URL: `https://apihub.agnes-ai.com/v1` - Optional base URL override: `AGNES_BASE_URL` - API key env var: `AGNES_API_KEY` - Auth header: `Authorization: Bearer $AGNES_API_KEY` - Official docs: `https://agnes-ai.com/en/docs/overview` - Platform: `https://platform.agnes-ai.com/` ## Regional Endpoint Routing Choose the service route before configuring a client. Do not automatically rotate an API key across services after an authentication or quota error. | Service route | Base URL | When to use it | | --- | --- | --- | | International service (primary) | `https://apihub.agnes-ai.com/v1` | Default route for the international service | | International service (alternate) | `https://apihub.agnes-ai.cn/v1` | Only when the primary route has a network, DNS, TLS, or connection-timeout failure | | China service | `https://api.agnes-ai.cn/v1` | Use for the China service | If the base URL cannot connect, confirm `/v1` is present, identify the intended service, and test at most one minimal alternate-route request. Do not switch routes to solve `400`, `401`, `403`, `422`, or `429` responses. ## Model Selection Use these defaults unless the user specifies a different model: | Workflow | Model | Endpoint | | --- | --- | --- | | Chat, coding, reasoning, tools, streaming, vision input | `agnes-2.5-flash` | `POST /v1/chat/completions` | | Image generation, editing, and multi-image composition | `agnes-image-2.5-flash` | `POST /v1/images/generations` | | Text-to-video, keyframes, and multimodal references | `agnes-video-2.5` | `POST /v1/videos` | | Fast 720P text-to-video and reference generation | `agnes-video-2.5-flash` | `POST /v1/videos` | For detailed model notes, read `references/model_catalog.md`.