Tessary
Tessary malformed output
malformed_output is one of Tessary's built-in classifiers, a deterministic one: checks that pass or fail, no judgment involved. When a call site declares a structured output, Tessary captures its schema and validates each output against it.
It catches exactly the failure your calling code would hit in production: output that doesn't parse, or parses but violates the declared shape. Every firing is high confidence, since the output either matches the schema or it doesn't, and detections carry the violation messages as evidence. Where validation isn't meaningful, it stays quiet. Detections go to triage, where the ones ruled real issues become cases.
6 questions
Answered, plainly.
Do I still need this classifier if my model's structured output mode is already on?Usually, yes. Structured output modes cut failures, they don't guarantee zero, and Tessary validates every call site regardless of what's enabled upstream.answer →Does Tessary use an LLM to check for malformed output?No. Validating an output against its declared schema is a deterministic check, a pass or fail with no model call, cheap enough to run on every trace.answer →Is malformed output invalid JSON, or just the wrong shape?Both. It fires when output fails to parse at all, and when output parses fine but violates the schema: wrong type, a missing field, or an extra one.answer →What is Tessary's malformed_output classifier?A deterministic classifier that validates every structured output against the schema its call site declared, and flags any that don't match.answer →Why doesn't malformed_output fire on every call site?Because it only has a schema to check where one was declared. A call site returning free text has no shape for the classifier to violate.answer →Why is my agent's output failing the schema it declared?Usually it's not garbled output, it's output that parses but drifts from the shape: a missing field, the wrong type, or prose wrapped around the JSON.answer →