AISchemaGate docs
Validate the JSON your models return against a contract, before your code acts on it. Start with the playbook; keep the SDK reference open while you build.
LLM output validation playbook
Author an output contract, validate one response, auto-repair, mask prompts, and provider quickstarts for OpenAI, Anthropic, vLLM, Ollama, LangChain, CrewAI and MCP.
aischemagate.dev/llm-output-validation.md
LLMGate SDK reference
Python and TypeScript side by side: constructor options, completion, message, validate_only, auto-repair, mask_prompt, errors.
aischemagate.dev/llm-sdk-reference.md
Using a coding agent?
Paste aischemagate.dev/llms.txt into Claude, Cursor or Codex. It links the raw markdown of both docs.
Ingesting data rather than validating model output? See the DataContractGate integration playbook.