Whatever message this page gives is out now! Go check it out!
ColdFusion AI framework schema reference
Last update:
May 18, 2026
Look up supported schema types, validation error messages, and provider feature support for the ColdFusion AI framework JSON schema guardrail system.
Supported primitive types
| Type string | Accepts | ColdFusion equivalent |
|---|---|---|
"String" | Any text value | isSimpleValue() |
"Number" | Integer or decimal | isNumeric() |
"Boolean" | true / false | isBoolean() |
Type strings are case-insensitive.
"String", "string", "STRING", and "sTrInG" are all equivalent.Optional fields
Append a
? suffix to mark a field as optional. The model includes it when the information is available and omits it when it is not. The ? suffix works on nested struct fields too.SCHEMA: {
"city": "String", // required
"country": "String", // required
"population": "Number", // required
"mayor": "String?", // optional
"elevation": "Number?" // optional
}Nested object support
Nest a CF struct inside the schema to describe a nested JSON object. Nesting up to five levels deep is verified to work. Anthropic has the best reliability for three or more levels of nesting.
SCHEMA: {
"company": "String",
"headquarters": {
"country": "String",
"office": {
"floor": "Number",
"room": "String"
}
}
}Array support
Wrap a single-element CF array around a type string or struct to describe an array field. An array schema value must contain exactly one element — an empty array or multi-element array throws before any API call is made.
// Array of primitives
SCHEMA: { "team": "String", "members": ["String"] }
// Array of objects
SCHEMA: {
"class": "String",
"students": [{ "name": "String", "grade": "Number", "pass": "Boolean" }]
}Unsupported types
| Rejected type | Use instead |
|---|---|
"Date" | "String" (specify the format in the prompt) |
"Object" | Nested struct { ... } |
"Double" | "Number" |
"Float" | "Number" |
"Int" | "Number" |
Schema validation errors
| Error message (partial) | Cause | Fix |
|---|---|---|
"cannot be null or empty" | SCHEMA: {} — empty struct | Add at least one field to the schema |
"valid ColdFusion struct" | SCHEMA is not a struct | Pass a CF struct |
"SYSTEMMESSAGE or USERMESSAGE" | Neither message key present | Add USERMESSAGE or a system message |
"did not complete" | Unsupported type name | Replace with "String", "Number", or "Boolean" |
"exactly one element" | Array schema has 0 or 2+ elements | Use exactly one element: ["String"] or [{...}] |
Runtime response anomalies
| Symptom | Check | Recovery |
|---|---|---|
response.message is empty | len(trim(text)) == 0 | Retry or return a default value |
| Response is not JSON | !isJSON(text) | Log the raw text and return an error |
| Required field missing | !structKeyExists(parsed, field) | Flag for review; use a default value |
| Wrong type for a field | !isNumeric(parsed.age) | Cast the value or flag it as invalid |
| Array is empty | arrayLen(parsed.items) == 0 | Retry with a more explicit prompt hint |
Provider compatibility matrix
| Feature | Anthropic | OpenAI | Azure OpenAI | Gemini | Mistral | Ollama |
|---|---|---|---|---|---|---|
| Flat schema | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
| Nested struct | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
| String array | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
| Number array | ✓ | ✓ | ✓ | ✓ | ✓ | — |
| Object array | ✓ | ✓ | ✓ | ✓ | ✓ | — |
Optional fields (?) | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
| System message | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
| Deep nesting (5+ levels) | ✓ | — | — | — | — | — |
temperature: 0 | — | ✓ | ✓ | — | ✓ | — |
Requires ResponseFormat | — | — | — | — | ✓ | ✓ |
Requires BASEURL | — | — | ✓ | — | — | ✓ |
The schema validation rules, type rejection and array arity checks, apply uniformly regardless of provider. For provider setup code, see configure AI providers with schema guardrails.
We're glad. Tell us how this page helped.
We're sorry. Can you tell us what didn't work for you?
Share additional feedback to help us improve.
0/255
|
