Image3 Prompt MCP
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@Image3 Prompt MCP Find me cinematic portrait prompts in the photorealistic category"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Image3 Prompt MCP
Image3 Prompt MCP is a read-only MCP server for GPT Image 2 prompt research. It lets Codex, Claude Code, Cursor, OpenClaw, and other MCP hosts search the public Image3 prompt gallery, inspect complete prompts, list categories, and turn rough ideas into structured prompt briefs.
Prompt gallery: https://image3.org/prompts/gpt-image-2
Prompt Architect docs: https://image3.org/docs/gpt-image-2-prompt-architect
Generator: https://image3.org/models/gpt-image-2
OpenClaw skill source: openclaw/gpt-image-2-prompt-architect/SKILL.md
Chinese README: README.zh-CN.md
The server is intentionally read-only. It does not generate images, spend credits, create accounts, or require an API key.
Tools
Tool | Purpose |
| Search Image3 prompts by query, category, language, and featured status. |
| Fetch one complete sanitized prompt by slug or id. |
| List prompt categories and counts. |
| Turn a rough idea into a deterministic structured prompt brief. |
Every prompt result includes its canonical Image3 source page.
Related MCP server: Durable MCP Tripleshot
Install from GitHub
{
"mcpServers": {
"image3": {
"command": "npx",
"args": ["-y", "github:gpt-img-2/image3-prompt-mcp"]
}
}
}Codex:
codex mcp add image3 -- npx -y github:gpt-img-2/image3-prompt-mcpOpenClaw:
openclaw mcp add image3 \
--command npx \
--arg -y \
--arg github:gpt-img-2/image3-prompt-mcp \
--include 'search_prompts,get_prompt,list_categories,build_prompt_brief'
openclaw mcp doctor image3 --probeOptional environment variables
Variable | Default | Purpose |
|
| Image3 prompt API origin. |
|
| Origin used for canonical source links. |
|
| Prompt API timeout from 1,000 to 60,000 ms. |
Development
pnpm install
pnpm validate
pnpm smoke:livepnpm validate runs type checking, unit tests, a build, and a local MCP handshake/tool-call smoke test. pnpm smoke:live additionally verifies the public Image3 prompt API.
Available Tools
4 toolsbuild_prompt_briefBuild a GPT Image 2 prompt briefARead-onlyIdempotent
Turn a rough image idea into a structured, deterministic GPT Image 2 brief with constraints and revision moves. This does not generate an image.
Inputs: idea (required), style, useCase, exactText, +2 more.
get_promptGet an Image3 promptARead-onlyIdempotent
Get one complete sanitized prompt and its media from Image3 by slug or id. Use an identifier returned by search_prompts.
Inputs: slugOrId (required).
list_categoriesList Image3 prompt categoriesARead-onlyIdempotent
List Image3 prompt categories with counts for an optional language. Read-only and free.
Inputs: language.
search_promptsSearch Image3 promptsARead-onlyIdempotent
Search the Image3 GPT Image 2 prompt gallery. Returns concise previews, media, categories, and canonical Image3 source pages. Read-only and free.
Inputs: limit, query, category, language, +1 more.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 4 tools
Each tool has a clearly distinct role: searching gallery previews, retrieving a full prompt, listing categories, and building a prompt brief. The overlap between search_prompts and get_prompt is well-handled because search returns concise previews while get_prompt requires a returned identifier and returns the complete prompt.
All tool names follow the same verb_noun snake_case pattern: search_prompts, get_prompt, list_categories, and build_prompt_brief. The verbs are specific and consistent with each operation.
Four tools is well-scoped for a prompt gallery MCP. Each tool earns its place and there is no unnecessary duplication or excessive surface area.
The server covers the core workflow of discovering prompts, retrieving full prompt details, browsing categories, and constructing a new prompt brief. Read-only gallery operations plus the brief-building step form a complete, sensible workflow with no obvious dead ends.
Maintenance
Related MCP Connectors
Search attributed image prompts, retrieve full templates, and browse GPT Image Hub categories.
Your prompt library in any MCP client: list, fetch, save and version prompt templates.
Self-hosted AI prompt library: prompts, collections, tags, teams, chains. 29 MCP tools for agents.
Team prompt library with version control and reviews. Save, search and reuse prompts via MCP.
Related MCP Servers
- AlicenseNot gradedqualityDmaintenanceProvides tools for generating and editing images using OpenAI's gpt-image-1 model via an MCP interface, enabling AI assistants to create and modify images based on text prompts.15Apache 2.0
- FlicenseNot gradedqualityDmaintenanceEnables users to search, browse, and retrieve AI prompts from the Tripleshot public API. Provides access to community prompts, trending content, and template rendering capabilities through a durable MCP server interface.-
- AlicenseAqualityDmaintenanceOpen-source MCP server for AI image and video creation, enabling prompt library search, prompt enhancement, and media generation from Claude Code, Cursor, and other MCP hosts.8MIT
- AlicenseAqualityDmaintenanceReads saved Keep My Prompts library and fetches prompts by name from any MCP client, read-only.3MIT