GPT Image 2 Polza MCP Server
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| POLZA_BASE_URL | No | Базовый URL API; по умолчанию https://polza.ai/api | https://polza.ai/api |
| GPT_IMAGE_MODEL | No | Model tier; по умолчанию gpt-image-2 | gpt-image-2 |
| IMAGE_OUTPUT_DIR | No | Папка результатов; по умолчанию ~/gpt-image-2-images | ~/gpt-image-2-images |
| POLZA_AI_API_KEY | Yes | API-ключ аккаунта Polza | |
| RETURN_FULL_IMAGE | No | Возвращать полное изображение вместо превью | |
| POLZA_EXTERNAL_USER_ID | No | Внешний ID пользователя для Polza | |
| POLZA_POLL_TIMEOUT_SECONDS | No | Максимальное время ожидания результата | |
| POLZA_POLL_INTERVAL_SECONDS | No | Интервал проверки статуса генерации |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| experimental | {
"tasks": {
"list": {},
"cancel": {},
"requests": {
"tools": {
"call": {}
},
"prompts": {
"get": {}
},
"resources": {
"read": {}
}
}
}
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| generate_imageA | Generate new images or edit existing images using natural language instructions. Supports multiple input modes:
Automatically detects mode based on parameters or can be explicitly controlled. Input images are read from the local filesystem to avoid massive token usage. Returns both MCP image content blocks and structured JSON with metadata. |
| upload_fileA | Upload a local file through the Polza Storage API and return its URI & metadata. Useful when the image is larger than 20MB or reused across prompts. |
| show_output_statsA | Show statistics about the output directory and recently generated images. |
| maintenanceB | Perform maintenance operations following workflows.md patterns. Available operations:
|
| fetch_generationD | – |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
| photorealistic_shot | Generate a prompt for high-quality photorealistic images. |
| logo_text | Generate a prompt for logo creation with accurate text rendering. |
| product_shot | Generate a prompt for studio product photography. |
| sticker_flat | Generate a prompt for flat/kawaii style stickers. |
| iterative_edit_instruction | Generate an instruction for precise image editing. |
| composition_and_style_transfer | Generate an instruction for style transfer and composition blending. |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
| prompt_templates_catalog | A compact catalog of prompt templates (same schemas as the @mcp.prompt items). |
| list_operations | List all tracked operations. Returns: Dict with list of operations and summary statistics |
TDQS
Scored across 5 tools
generate_image, upload_file, show_output_stats, and maintenance each have clear and separate purposes. fetch_generation is the only ambiguous tool because it lacks a description and may be confused with retrieval or generation workflows.
Four tools follow a verb_noun pattern (generate_image, upload_file, show_output_stats, fetch_generation), but maintenance is a bare noun instead of something like perform_maintenance. The pattern is still readable and mostly predictable.
Five tools is well-scoped for an image generation server: one core generation tool, upload support, stats, fetch, and maintenance. Each tool covers a distinct functional area without redundancy.
The set covers the primary generation/editing, upload, retrieval, stats, and cleanup workflows. Minor gaps include no explicit per-file delete/list operations and fetch_generation's undocumented behavior, but agents can generally work around them via maintenance and cleanup.