Documentation

ValGuard.ai sits between your app and LLM providers. Validators run on every completion — 160 deterministic rules across Core, Compliance, and Code packs. Growth+ Provider vault custom endpoints extend that to privately hosted inference (including vLLM, Ollama, LiteLLM, and internal corporate models).

API Reference

Full endpoint index, payload cheat-sheet, auth model, and OpenAPI 3.2 spec

Getting Started

Five steps from signup to first validated response + top templates

Tutorials

23 step-by-step guides: playbooks, validation, RAG, deployment, security, and monitoring

Quickstart

Python, Node.js, Go, C#, curl, Zapier

Playbooks

51 production-ready workflows with X-VG-Flow, billing, and SDK examples

Supported providers

28 backends, provider/model routing, and Growth+ custom URLs for local or on-prem stacks

Templates

116 curated validator packs for invoice extraction, RAG QA, PII detection, and more

Error reference

HTTP status codes, validation headers, and remediation guidance

Cost control

Budgets, playbook routes, smart-fallback, model prices, priced vs estimated labels

Validators catalog

160 built-in rules with defaults and categories

Playground

Test validators without an LLM call

FAQ

Plans, billing, packs, deployment

Minimal curl

Replace with your API key from the dashboard. See full quickstart.

export VG_PROXY="https://api.valguard.ai"
export VG_API_KEY="vg_live_YOUR_KEY"
export VG_AGENT="default"

curl -s "$VG_PROXY/v1/chat/completions" \
  -H "Authorization: Bearer $VG_API_KEY" \
  -H "X-VG-Agent: $VG_AGENT" \
  -H "Content-Type: application/json" \
  -d '{"model":"openai/gpt-4o-mini","messages":[{"role":"user","content":"Say hi"}]}'

Inbound prompt validation

Inbound prompt validation scans user messages before they reach the LLM. It is off by default. Enable it on an agent when you need to catch PII, secrets, or injection patterns in incoming prompts.

Shadow — log matches but still forward to the model.
Block — reject the request before upstream call.

Presets and plan limits are configured per agent under Agent settings. Outbound response validators are separate — see the validators catalog.

Supported providers

Every plan includes all 28 built-in LLM backends — set the model to provider/model (for example openai/gpt-4o-mini) and pass your provider key via X-Provider-Key or the Provider vault on Growth+.

See the full provider list for example models, region configuration, and custom on-prem endpoints.