How to Connect Stability AI to Microsoft Copilot via MCP

💡 No install? Use cloud.anythingmcp.com directly. Sign in, click Connectors → Stability AI, paste your credentials, mint an MCP API key and you're done. No Docker, no git clone, no local server.

Stability AI + Microsoft Copilot

Create and edit pictures with Stability AI: generate images with Stable Image Ultra, Core and Stable Diffusion 3.5, upscale, remove or replace backgrounds, erase, inpaint, outpaint, search and replace objects, guide generation with sketches, structure and style references, and create music with Stable Audio 3. Pictures come back as images; it also reads the credit balance. Paid per generation in credits.

Prerequisites

See the full setup instructions baked into the connector (visible in the in-app store when you select the connector). The required environment variables for this connector are:

STABILITY_API_KEY

Step 1 — Get credentials

Setting up takes about 3 minutes. You need a Stability AI account with credits; new accounts get 25 free credits.

1. Create an account at https://platform.stability.ai and buy credits under Billing if needed.
2. Open API keys (https://platform.stability.ai/dashboard/api-keys), copy a key (it starts with sk-) and paste it into STABILITY_API_KEY. It is sent as Authorization: Bearer.
3. Install the connector and run stability_ai_get_balance: it shows the credit balance and confirms the key. The key acts for the default organization of its user, and credits are taken from that organization.

Costs: 1 credit = $0.01, and failed generations are free. Every generation, edit, upscale and audio tool costs credits:

• Create from text: stability_ai_generate_image_core 3, stability_ai_generate_image_ultra 8, stability_ai_generate_image_sd3 6.5 (sd3.5-large), 4 (sd3.5-large-turbo, the default here), 3.5 (sd3.5-medium) or 2.5 (sd3.5-flash).
• Upscale: stability_ai_upscale_fast 2, stability_ai_upscale_conservative 40, stability_ai_upscale_creative 60.
• Edit: stability_ai_erase, stability_ai_inpaint, stability_ai_search_and_replace, stability_ai_search_and_recolor and stability_ai_remove_background 5 each, stability_ai_outpaint 4, stability_ai_replace_background_and_relight 8.
• Guide with a picture: stability_ai_control_sketch, stability_ai_control_structure and stability_ai_control_style 5 each, stability_ai_control_style_transfer 8.
• Audio: stability_ai_generate_audio 8.

Free: stability_ai_get_balance, stability_ai_get_account, stability_ai_list_engines, stability_ai_get_result and stability_ai_get_audio_result. Before every paid call the AI tells you the cost and waits for your yes; when in doubt it checks the balance first.

What it creates and edits:

• Pictures from text: Core is fast and the cheapest default; Ultra gives the best quality and photorealism (about 1 megapixel) and can also start from a picture with strength; Stable Diffusion 3.5 offers Large, Large Turbo, Medium and Flash, also image to image with mode, image_url and strength. Aspect ratios go from 21:9 to 9:21.
• Higher resolution: the fast upscaler enlarges 4 times without reinterpreting the picture, the conservative one goes to about 4 megapixels and stays faithful, the creative one goes up to 4K and adds detail guided by a prompt.
• Edits: remove the background (transparent PNG or WebP), erase objects or blemishes with a mask, repaint a masked area (inpaint), extend the canvas in any direction (outpaint), find an object by words and replace or recolour it without a mask, and put a subject on a new background with matching light.
• Guided generation: turn a sketch into a finished picture, keep the structure of an input picture, copy the style of a reference, or transfer the style of one picture onto another.
• Music and sound effects with Stable Audio 3, from 1 to 380 seconds, as MP3 or WAV.
• The account: credit balance, the user with its organizations, and the legacy v1 engines (SDXL and older; the current services are not listed there).

Input pictures: every *_url parameter is a public https link to a JPEG, PNG or WebP file of up to 10 MB. AnythingMCP downloads it and uploads it to Stability AI as a file, so a picture you only pasted into the chat has to be put online first. In a mask, white marks the area to change; without a mask the transparent areas of the picture are used.

Asynchronous jobs: stability_ai_upscale_creative, stability_ai_replace_background_and_relight and stability_ai_generate_audio return a 64-character id at once. The AI calls stability_ai_get_result (pictures) or stability_ai_get_audio_result (audio) with it after about 10 seconds, and again while status is in-progress. Results stay available for 24 hours and only for the same key.

Data formats: finished pictures are attached to the tool result as images you can see, audio as a file. Picture tools also return finish_reason and seed; the same seed with the same prompt reproduces a picture. Pictures default to JPEG to stay small; ask for png or webp when you need transparency. Prompts work best in English and describe subject, setting, style, colours and lighting; what to avoid goes into negative_prompt, and style_preset (photographic, cinematic, anime, digital-art, 3d-model, ...) nudges the look.

Paging: there is nothing to page through. Each call returns one picture, one audio job or one account record.

Switched off at install: no tool. Every tool is enabled; the paid ones are guarded by the cost confirmation.

Limits: 150 requests in 10 seconds; above that, requests are blocked for 60 seconds. Prompts can be up to 10,000 characters.

Common errors: 400 lists the wrong parameters in errors. 401 means a bad key. 402 means not enough credits. 403 means the prompt or picture was flagged by content moderation. 413 means the upload is too large. 422 means the input cannot be processed, for example a picture size out of range or an unsupported language. 429 is the rate limit. finish_reason CONTENT_FILTERED means the output broke the content policy and was blurred: rephrase the prompt. Every error carries an id to quote to Stability support.

The connector does not generate 3D models (Stable Fast 3D, SPAR3D), does not do audio to audio or audio inpainting, and does not use the Stable Audio 2 endpoints or the legacy v1 SDXL generation.

Step 2 — Install the adapter

curl -fsSL https://raw.githubusercontent.com/HelpCode-ai/anythingmcp/main/docker-compose.quickstart.yml -o docker-compose.yml
printf 'JWT_SECRET=%s\nENCRYPTION_KEY=%s\n' "$(openssl rand -hex 32)" "$(openssl rand -hex 32)" > .env
docker compose up -d

Step 3 — Add the connector in Microsoft Copilot

Your server URL: in AnythingMCP, open MCP Servers → the server this connector is on, and copy its URL (https://cloud.anythingmcp.com/mcp/…). Use it wherever this guide shows YOUR_SERVER_ID.

Microsoft Copilot uses MCP servers through agents built in Copilot Studio: you add AnythingMCP to an agent as a tool, then publish the agent to Microsoft 365 Copilot and Teams. Copilot Studio connects from Microsoft's cloud over Streamable HTTP, which AnythingMCP supports.

1. In Copilot Studio, open your agent (or create one), go to Tools and choose Add a tool → New tool → Model Context Protocol.
2. Enter a Server name (for example AnythingMCP), a Server description that says what the connector does, and the Server URL https://cloud.anythingmcp.com/mcp/YOUR_SERVER_ID. Copilot Studio uses the description to decide when to call the server.
3. Under authentication, choose API key, set Type to Header and the header name to X-API-Key, then select Create.
4. In Add tool, choose Create a new connection, paste an MCP API key from AnythingMCP (MCP Servers → your server → API keys) and select Add to agent.
5. Try it in the agent's Test pane: ask for something that needs the Stability AI tools.
6. Select Publish, then Channels → Microsoft 365 and Microsoft Teams, keep Make agent available in Microsoft Copilot selected and choose Add channel. See agent in Teams installs it for you; in Microsoft 365 Copilot, type @ and pick the agent.

Prefer per-user sign-in? Choose OAuth 2.0 → Dynamic discovery instead of an API key: each user then signs in to AnythingMCP the first time the agent calls it.

Self-hosting? Use your instance's public HTTPS address instead of https://cloud.anythingmcp.com. Copilot Studio connects from Microsoft's cloud, so localhost does not work.

Available tools

Example prompts

• "How many Stability AI credits are left on the account?"
• "Create a photorealistic picture of a lighthouse on a rocky coast at sunset in 16:9 with Stable Image Ultra. Tell me the cost first."
• "Remove the background from https://example.com/product.jpg and give me a transparent PNG."
• "In https://example.com/street.jpg, replace the red car with a blue bicycle."
• "Make a 30-second calm lo-fi piano track at 80 BPM for a podcast intro."

FAQ

Which Copilot is this for? Microsoft 365 Copilot at work, through agents built in Copilot Studio. The Microsoft Copilot app for personal accounts has no option to add your own MCP server.

What do I need? A work account with access to Copilot Studio (a license or a trial) to build the agent. People who use it in Microsoft 365 Copilot or Teams need access to the agent: share it with them, or ask an admin to publish it to the organization. If your organization applies data policies to Power Platform connectors, they cover MCP servers too.

The agent doesn't call the tools. Check that the header name is exactly X-API-Key and that the key belongs to the same MCP server as the URL, and make the server description specific: Copilot Studio uses it to decide when to call the server. Publish the agent again after each change; in an open chat, type "start over" to get the new version.

Next steps

• Connector landing page
• Connect to Claude
• Connect to ChatGPT
• Connect to Cursor
• Connect to Gemini
• Connect to Meta Muse