LLM.txt Documentation
Structured markdown documentation optimized for AI assistants and agent runtimes
What is LLM.txt?
AI-optimized documentation for seamless integration
LLM.txt is a structured markdown file designed specifically for AI assistants like Claude, GPT, and other large language models. Instead of navigating complex documentation, you can paste the LLM.txt content into your AI conversation or runtime and get instant, accurate help with API integration and model selection.
Complete API Reference
All endpoints, schemas, and response formats in one place
Agent Reasoning Context
Useful for model selection, endpoint planning, and deciding when to poll vs webhook
Tooling Companion
Pairs with OpenAPI and per-model schema routes for SDK, CLI, and MCP-style tools
How to Use LLM.txt
Three simple ways to leverage AI-powered documentation
Copy and Paste
Open the LLM.txt file, copy its contents, and paste it into your conversation with Claude, GPT, or any AI assistant. Then ask questions like "Which model should I use?", "What schema does this model accept?", or "Show me Python code to create a video."
https://pixeldojo.ai/llm.txtFetch Programmatically
Use the LLM.txt as context in your AI-powered applications or alongside the schema/control-plane endpoints:
import requests
# Fetch the documentation
llm_docs = requests.get("https://pixeldojo.ai/llm.txt").text
# Use with OpenAI
from openai import OpenAI
client = OpenAI()
response = client.chat.completions.create(
model="gpt-4",
messages=[
{"role": "system", "content": f"You are a helpful assistant..."},
{"role": "user", "content": "How do I generate an image?"}
]
)Per-Model Documentation
Each model has its own LLM.txt file with focused documentation. Use this when you only need info about a specific model:
/llm/{model-id}.txtExample: /llm/flux-1.txt, /llm/kling-v2-5.txt
Per-Model LLM Documentation
Click any model to view its dedicated LLM.txt file
Recommended Agent Tooling Shape
How LLM docs fit into SDK, CLI, and MCP-style integrations
Use /llm.txt and per-model docs as the reasoning layer, then pair them with /api/openapi and /api/v1/models/{apiId}/schema for exact request construction.
This pattern works well for three install surfaces: generated SDKs with helper docs, CLIs that need human-readable guidance, and MCP-style tools that let an agent call list_models, get_model_schema, generate_media, and list_jobs.