AI & Machine Learning
aile.sh
An MCP server that provides tools for discovering and using AI models for chat, image generation/editing, and text-to-speech.
ENDPOINT 1
https://api.aile.sh/mcp
MCP server metadata
- Name
- aile
- Version
- 0.1.0
Aile: AI models and other agents' tools, paid per call. You are connected without an account, so you pay each call from your own wallet (x402). USE A MODEL (an LLM, or image/audio/video/search): find_models -> use_model (media: generate_image, text_to_speech, ...). Priced per token (on max_tokens) or per unit. USE ANOTHER AGENT'S TOOL: find_agent_tools -> use_agent_tool. A flat price per call. A paid call returns its price as an x402 challenge; sign it and call again with the PaymentPayload in _meta["x402/payment"]. Where MPP is on, the same result carries MPP challenges in _meta["org.paymentauth/payment-required"]; pay with the credential in _meta["org.paymentauth/credential"]. Prompts: use_models, use_agent_tools.
Known tools 13
find_models[Use a model] Find AI models you can use, for one kind of output: `chat` (LLMs, the default), or image, audio, music, video, search, scrape, rerank, moderation, ocr.
Inferred read-onlyuse_model[Use a model] Call an AI model (an LLM) once: send a prompt or a conversation, get its reply.
Potential side effectsgenerate_image[Use a model] Generate images from a prompt (FLUX, Gemini, GPT image and more).
Inferred read-onlyedit_image[Use a model] Edit an image from a prompt and one or more input images (https URLs or data URIs).
Potential side effectstext_to_speech[Use a model] Speak text aloud (ElevenLabs, OpenAI, Fish, MiniMax and more).
Inferred read-onlytranscribe_audio[Use a model] Transcribe audio to text (Whisper, Deepgram, ElevenLabs Scribe and more), or translate it to English.
Inferred read-onlygenerate_music[Use a model] Generate a music track from a prompt (one-shot models such as Lyria clip; gemini/lyria-3.5, Eleven Music and render-job models such as kie/Suno need a held POST /v1/music/generations).
Potential side effectsweb_search[Use a model] Search the web (Brave, Firecrawl, ScrapingDog, Serper, SearchAPI.io, Perplexity).
Inferred read-onlyscrape_url[Use a model] Fetch one URL and convert it to markdown or HTML (Firecrawl, ScrapingDog, ScrapingBee).
Inferred read-onlygenerate_video[Use a model] Start a video render from a prompt (Veo, Seedance, Wan and more).
Inferred read-onlyget_video[Use a model] Poll a video job from generate_video: its status, and the content URL once completed.
Inferred read-onlyfind_agent_tools[Use an agent's tool] Find tools OTHER AGENTS offer: one row per listing, with its lenderId, what it does, its flat price per call, and the names of its tools.
Inferred read-onlyuse_agent_tool[Use an agent's tool] Call one tool another agent offers, at its flat price per call (find it with find_agent_tools; to call an AI model instead, see use_model).
Inferred read-onlyCONNECT WITH APPROVAL
Client installation
Review this server and its permissions before adding it. Secret placeholders must be set locally.
Codex
~/.codex/config.toml
[mcp_servers.aile]
url = "https://api.aile.sh/mcp"
enabled = true
Claude Code
.mcp.json
{
"mcpServers": {
"aile": {
"type": "http",
"url": "https://api.aile.sh/mcp"
}
}
}
Claude Desktop
Settings → Connectors → Add custom connector
Name: aile
Remote MCP URL: https://api.aile.sh/mcp
Add this remote URL as a custom connector in Claude Desktop. Availability depends on the user plan and workspace policy.
Cursor
.cursor/mcp.json
{
"mcpServers": {
"aile": {
"url": "https://api.aile.sh/mcp"
}
}
}
Visual Studio Code
.vscode/mcp.json
Add to Visual Studio Code{
"servers": {
"aile": {
"type": "http",
"url": "https://api.aile.sh/mcp"
}
}
}
Generic MCP
Client-specific MCP configuration
{
"name": "aile",
"transport": "streamable-http",
"url": "https://api.aile.sh/mcp"
}
MCP Inspector
Run the official MCP Inspector locally and enter the indexed Streamable HTTP endpoint.
TRUST AND VERIFICATION EVIDENCE
Loading Trust v2 evidence…
Checking the associated registrable domain. The BuiltWith key remains server-side.
Evidence is source-attributed and does not guarantee that a third-party server is safe. Risk labels are conservative metadata heuristics.