OpenAI Automation
Automate OpenAI API operations -- generate responses with multimodal and structured output support, create embeddings, generate images, and list models via the Composio MCP integration.
Get the complete skill folder, including its scripts and reference files.
This workflow depends on an external MCP integration. Check the source for the server, account access and setup it requires.
When to use this skill
Use OpenAI Automation when an AI agent needs a reusable SKILL.md workflow for this job: Automate OpenAI API operations -- generate responses with multimodal and structured output support, create embeddings, generate images, and list models via the Composio MCP integration.
When not to use it
Skip OpenAI Automation when the task is outside the coding category, or when a more specific skill in this directory already covers the same workflow with clearer triggers.
How to install
Where will you use it?
mkdir -p ~/.claude/skills/openai-automation # Save SKILL.md to ~/.claude/skills/openai-automation/SKILL.md
- Claude Code personal install: place the skill and its bundled files at ~/.claude/skills/openai-automation/SKILL.md. For a shared project, commit the folder at .claude/skills/openai-automation/ instead.
- Current Claude Code detects SKILL.md changes in watched directories during the session. If you just created a top-level skills directory, run /reload-skills. Check that the skill is listed before testing its trigger.
- For Claude chat or Cowork in the Desktop app, manage uploaded skills in Customize > Skills; copying files into a local Claude Code folder is not a Desktop install. Other agents may use different install locations.
Full install guide for Claude, Cursor, and Codex →
Try a starter prompt
After installing the skill, paste this starter prompt into your agent to verify it triggers on a small task.
Use the OpenAI Automation skill for this task: Use OpenAI Automation when this documented workflow matches the task. Confirm any required inputs or files before making changes.
What supports this recommendation?
72/100 provisional content score. This measures documentation, practical detail and source signals. It is not a measured task success rate or a security certification.
- Documentation
- 77/100
- Practical detail
- 68/100
- Source evidence
- 63/100
- Source signals
- 80/100
Model compatibility evidence
- claudeinferred
The skill text mentions Claude or a closely associated term. This is keyword-derived compatibility; no hands-on model test is recorded.
- chatgptinferred
The skill text mentions ChatGPT or a closely associated term. This is keyword-derived compatibility; no hands-on model test is recorded.
- geminiuntested
No model-specific signal or recorded compatibility test was found.
- copilotuntested
No model-specific signal or recorded compatibility test was found.
- llamauntested
No model-specific signal or recorded compatibility test was found.
- perplexityuntested
No model-specific signal or recorded compatibility test was found.
- mistraluntested
No model-specific signal or recorded compatibility test was found.
- grokuntested
No model-specific signal or recorded compatibility test was found.
Inferred compatibility is based on content signals. Check the app, its available tools and setup requirements before running a skill.
What this skill does
# OpenAI Automation
Automate your OpenAI API workflows -- generate text with the Responses API (including multimodal image+text inputs and structured JSON outputs), create embeddings for search and clustering, generate images with DALL-E and GPT Image models, and list available models.
**Toolkit docs:** [composio.dev/toolkits/openai](https://composio.dev/toolkits/openai)
---
## Setup
1. Add the Composio MCP server to your client: `https://rube.app/mcp` 2. Connect your OpenAI account when prompted (API key authentication) 3. Start using the workflows below
---
## Core Workflows
### 1. Generate a Response (Text, Multimodal, Structured)
Use `OPENAI_CREATE_RESPONSE` for one-shot model responses including text, image analysis, OCR, and structured JSON outputs.
``` Tool: OPENAI_CREATE_RESPONSE Inputs: - model: string (required) -- e.g., "gpt-5", "gpt-4o", "o3-mini" - input: string | array (required) Simple: "Explain quantum computing" Multimodal: [ { role: "user", content: [ { type: "input_text", text: "What is in this image?" }, { type: "input_image", image_url: { url: "https://..." } } ]} ] - temperature: number (0-2, optional -- not supported with reasoning models) - max_output_tokens: integer (optional) - reasoning: { effort: "none" | "minimal" | "low" | "medium" | "high" } - text: object (structured output config) - format: { type: "json_schema", name: "...", schema: {...}, strict: true } - tools: array (function, code_interpreter, file_search, web_search) - tool_choice: "auto" | "none" | "required" | { type: "function", function: { name: "..." } } - store: boolean (false to opt out of model distillation) - stream: boolean ```
**Structured output example:** Set `text.format` to `{ type: "json_schema", name: "person", schema: { type: "object", properties: { name: { type: "string" }, age: { type: "integer" } }, required: ["name", "age"], additionalProperties: false }, strict: true }`.
### 2. Create Embeddings
Use `OPENAI_CREATE_EMBEDDINGS` for vector search, clustering, recommendations, and RAG pipelines.
``` Tool: OPENAI_CREATE_EMBEDDINGS Inputs: - input: string | string[] | int[] | int[][] (required) -- max 8192 tokens, max 2048 items - model: string (required) -- "text-embedding-3-small", "text-embedding-3-large", "text-embedding-ada-002" - dimensions: integer (optional, only for text-embedding-3 and later) - encoding_format: "float" | "base64" (default "float") - user: string (optional, end-user ID for abuse monitoring) ```
### 3. Generate Images
Use `OPENAI_CREATE_IMAGE` to create images from text prompts using GPT Image or DALL-E models.
``` Tool: OPENAI_CREATE_IMAGE Inputs: - model: string (required) -- "gpt-image-1", "gpt-image-1.5", "dall-e-3", "dall-e-2" - prompt: string (required) -- max 32000 chars (GPT Image), 4000 (DALL-E 3), 1000 (DALL-E 2) - size: "1024x1024" | "1536x1024" | "1024x1536" | "auto" | "256x256" | "512x512" | "1792x1024" | "1024x1792" - quality: "standard" | "hd" | "auto" | "high" | "medium" | "low" - n: integer (1-10; DALL-E 3 supports n=1 only) - background: "transparent" | "opaque" | "auto" (GPT Image models only) - style: "vivid" | "natural" (DALL-E 3 only) - user: string (optional) ```
### 4. List Available Models
Use `OPENAI_LIST_MODELS` to discover which models are accessible with your API key.
``` Tool: OPENAI_LIST_MODELS Inputs: (none) ```
---
## Known Pitfalls
| Pitfall | Detail | |---------|--------| | DALL-E deprecation | DALL-E 2 and DALL-E 3 are deprecated and will stop being supported on 05/12/2026. Prefer GPT Image models. | | DALL-E 3 single image only | `OPENAI_CREATE_IMAGE` with DALL-E 3 only supports `n=1`. Use GPT Image models or DALL-E 2 for multiple images. | | Token limits for embeddings | Input must not exceed 8192 tokens per item and 2048 items per batch for embedding models. | | Reasoning model restrictions | `temperature` and `top_p` are not supported with reasoning models (o3-mini, etc.). Use `reasoning.effort` instead. | | Structured output strict mode | When `strict: true` in json_schema format, ALL schema properties must be listed in the `required` array. | | Prompt length varies by model | Image prompt max lengths differ: 32000 (GPT Image), 4000 (DALL-E 3), 1000 (DALL-E 2). |
---
## Quick Reference
| Tool Slug | Description | |-----------|-------------| | `OPENAI_CREATE_RESPONSE` | Generate text/multimodal responses with structured output support | | `OPENAI_CREATE_EMBEDDINGS` | Create text embeddings for search, clustering, and RAG | | `OPENAI_CREATE_IMAGE` | Generate images from text prompts | | `OPENAI_LIST_MODELS` | List all models available to your API key |
---
*Powered by [Composio](https://composio.dev)*
Intended uses
- Use OpenAI Automation when this documented workflow matches the task.
Implementation notes
- Review the source instructions and adapt inputs before running the workflow.
Try it, then tell us how it went
A copied prompt does not tell us whether the task worked. Report your result after checking the output. We collect your tool and outcome, never your files or prompt.