produce — Codex skill: what it does, how to install
Automates visual asset generation and reviews on a Codex interactive board. It creates ad directions, product scenes, and style concepts.
npx skills add openai/plugins --skill produce -a codexWhat produce does
This skill manages visual asset generation using the creative_production_board tool. It opens an interactive board session and registers placeholder IDs before rendering starts. If a single native image is needed, the coordinator runs Codex ImageGen directly. Requests with two or more images route through Desktop subagents.
Execution keeps a ceiling of 4 concurrent worker threads for native image generation. It retries failed or invalid files up to 2 total attempts per output. By default, the workflow produces 4-6 separate visual directions without building collages or inner mood boards.
The workflow pulls specific mode references for ads, camera shots, scenes, or charts only when relevant. Inventing unverified prices, claims, or text copy is explicitly prohibited.
What it helps with
- generate 4-6 distinct visual directions for marketing campaigns
- parallelize image creation across Desktop subagents
- explore camera angles and product scenes on an interactive board
- maintain exact logos and packaging details across generated tiles
Example prompts for the agent
- Create 4 ad concepts for our upcoming summer drink campaign
- Generate product scenes with this bottle on a marble countertop
- Explore new logo marks and typography directions for an apparel brand
How to install produce
1Skills CLI (Vercel)
npx skills add openai/plugins --skill produce -a codex2Manually: put the files in place
.agents/skills/produce/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add openai/plugins --skill produce -a claude-code2Manually: put the files in place
~/.claude/skills/produce/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/produce/
Manually: put the files in place
.cursor/skills/produce/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install openai/plugins produce2Manually: put the files in place
.github/skills/produce/SKILL.mdPlatform CLI
gemini skills install https://github.com/openai/plugins --path plugins/creative-production/skills/produceManually: put the files in place
.devin/skills/produce/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: Codex skills: how to install, where they live, how to call them
Questions about produce
How to install produce in Codex?
Run npx skills add openai/plugins --skill produce -a codex. The skill files will be placed into .agents/skills/produce/SKILL.md.
What does the produce skill do?
It orchestrates the creation of visual designs and marketing graphics inside an interactive board session. It assigns generation jobs and tracks revisions until completion.
How does produce handle concurrent image rendering?
Single native images run through built-in Codex ImageGen directly. For batches of two or more images, it delegates work to subagents while capping native worker threads at 4.
