mimo-3/mcp-imagenate

npx mcp-imagenate (or install globally: npm install -g mcp-imagenate)
Requires Node.js 20+ and at least one provider API key set in the environment.

Usage

• •Export API keys for providers you want: GEMINI_API_KEY / NANO_BANANA_API_KEY, OPENAI_API_KEY / GPT_IMAGE_API_KEY, BFL_API_KEY, REVE_API_KEY / REVE_API_TOKEN.
• •Claude Desktop: add a mcp-imagenate entry to claude_desktop_config.json (command npx, args [mcp-imagenate], env keys).
• •Set NANO_BANANA_OUTPUT_DIR to sandbox all output/input paths within one directory — recommended for production.
• •Give proxied calls at least a 120-second timeout — a generation costs ~150 credits (about $0.20) and typically takes 40-80 seconds.

Key features

• ✓Provider table: Gemini nano-banana-2 (gemini-3.1-flash-image-preview) and nano-banana-pro (gemini-3-pro-image-preview); OpenAI gpt-image-2; BFL FLUX klein-4b (flux-2-klein), pro-preview (flux-2-pro), max (flux-2-max); Reve (v2/image/create endpoint, latest is the only version alias).
• ✓One consistent generate_image tool across all providers with per-provider env key precedence.
• ✓Optional sandboxed output directory for production safety.
• ✓Published to npm as mcp-imagenate.

Best for

Agents that generate images from a choice of state-of-the-art image models with a single MCP tool and per-provider credentials.

Caveats

• ⚠At least one provider API key is required; each provider has cost/credit implications (a generation ~$0.20).
• ⚠Generations are slow (40-80s) — proxies/job runners need timeouts of 120s or more.
• ⚠Output files only land on the machine running the server (or the dir you designate); no cloud hosting or provisioning described in the README.