Agnes AI Skills

Official skill repository for Agnes AI model integration.

This repository provides a reusable Codex skill for integrating Agnes AI text, image, video, and agent models through the OpenAI-compatible API gateway.

Available Skills

Skill	Purpose
agnes-ai-models	Integrate Agnes AI text, image, video, and agent workflows.

Install From GitHub

Codex users can install the skill from this GitHub repository by targeting the skill directory:

python /path/to/install-skill-from-github.py \
  --repo AgnesAI-Labs/skills \
  --path agnes-ai-models

If your Codex environment supports URL-based installation, use:

python /path/to/install-skill-from-github.py \
  --url https://github.com/AgnesAI-Labs/skills/tree/main/agnes-ai-models

Restart Codex after installation so the skill is discovered.

Model Coverage

Model	Type	Primary use
agnes-2.5-flash	Text and vision-language	Chat, coding, reasoning, streaming, tools, and image understanding
agnes-image-2.1-flash	Image generation	Higher quality image generation and editing
agnes-video-v2.0	Video generation	Text-to-video, image-to-video, and video result polling

Agent Compatibility

This repository is packaged as a Codex Skill, but the skill content can also be used by other agent software such as OpenClaw, Hermes, Manus, and custom agents.

For non-Codex agents, use the files in agnes-ai-models as integration instructions:

• Read SKILL.md for the main workflow.
• Use references/model_catalog.md for model names, endpoints, RPM notes, and Token Plan notes.
• Use references/troubleshooting.md for common API errors.
• Use references/agent_compatibility.md for OpenClaw, Hermes, WorkBuddy, Manus, and generic agent setup guidance.

Agnes AI uses an OpenAI-compatible API gateway. Most agent frameworks that support custom OpenAI-compatible providers can use Agnes models by setting the Agnes base URL, API key, and model name.

Requirements

Before using the skill, each user must:

1. Register an Agnes Platform account: https://platform.agnes-ai.com/
2. Apply for an Agnes API key in the platform.
3. Store the key locally as an environment variable:

export AGNES_API_KEY="your_api_key_here"

Do not commit API keys, bearer tokens, .env files, screenshots containing secrets, or private customer data.

Regional Endpoint Routing

Select the service route before configuring a client:

Service route	Base URL
International service (primary)	https://apihub.agnes-ai.com/v1
International service (alternate)	https://apihub.agnes-ai.cn/v1
China service	https://api.agnes-ai.cn/v1

For the International service, start with the primary route. If it has a network, DNS, TLS, or connection-timeout failure, test the alternate route with one minimal request and keep the reachable route. For the China service, use the China service route. Do not switch routes to resolve 400, 401, 403, 422, or 429 responses.

Set the selected URL in an environment variable when supported:

export AGNES_BASE_URL="https://apihub.agnes-ai.com/v1"

Quick Start

Copy or install the agnes-ai-models skill into your Codex skills directory, or link the GitHub skill directory from another agent that supports external instructions. Then ask the agent to use it for Agnes AI API integration tasks.

Example prompt:

Use the Agnes AI Models skill to create a Python example for agnes-2.5-flash.

What The Skill Covers

• OpenAI-compatible chat completions
• Streaming responses
• Image generation with agnes-image-2.1-flash
• Video generation with agnes-video-v2.0 and video_id polling
• Tool-calling style agent workflows
• Common API errors and debugging checklists
• Token Plan and RPM reference notes

Official Links

• Website: https://agnes-ai.com/
• Docs: https://agnes-ai.com/doc/overview
• Platform: https://platform.agnes-ai.com/
• International API base URL (primary): https://apihub.agnes-ai.com/v1
• International API base URL (alternate): https://apihub.agnes-ai.cn/v1
• China API base URL: https://api.agnes-ai.cn/v1