OUTPUT
Response generation failures.
The OUTPUT module covers errors in generating and delivering agent responses. This includes structural formatting, response quality, and safety concerns.
Categories
FORMAT
Output formatting failures.
| Code | Severity | Description | Remediation |
|---|---|---|---|
OUTPUT.FORMAT.INVALID_JSON | 3 | Invalid JSON output | Use structured output parsing |
OUTPUT.FORMAT.SCHEMA_VIOLATION | 4 | Output violates schema | Validate against output schema |
OUTPUT.FORMAT.MALFORMED | 3 | Malformed output structure | Implement output validation |
OUTPUT.FORMAT.TRUNCATED | 2 | Output was truncated | Handle output length limits |
OUTPUT.FORMAT.ENCODING | 2 | Output encoding issues | Normalize output encoding |
QUALITY
Output quality failures.
| Code | Severity | Description | Remediation |
|---|---|---|---|
OUTPUT.QUALITY.INCOMPLETE | 3 | Incomplete response | Validate response completeness |
OUTPUT.QUALITY.IRRELEVANT | 3 | Response is irrelevant to query | Improve response relevance |
OUTPUT.QUALITY.VERBOSE | 2 | Overly verbose response | Implement response length optimization |
OUTPUT.QUALITY.UNCLEAR | 2 | Unclear or ambiguous response | Add clarity checks |
OUTPUT.QUALITY.REPETITIVE | 2 | Repetitive content in response | Detect and reduce repetition |
SAFETY
Output safety failures.
| Code | Severity | Description | Remediation |
|---|---|---|---|
OUTPUT.SAFETY.HARMFUL_CONTENT | 5 | Potentially harmful content | Implement content filtering |
OUTPUT.SAFETY.PII_LEAK | 5 | PII leaked in output | Implement PII detection and masking |
OUTPUT.SAFETY.BIAS | 4 | Biased content in output | Implement bias detection |
OUTPUT.SAFETY.POLICY_VIOLATION | 4 | Output violates policy | Add policy compliance checks |
OUTPUT.SAFETY.COPYRIGHT | 3 | Potential copyright violation | Implement copyright checks |
Danger
OUTPUT.SAFETY.HARMFUL_CONTENT and OUTPUT.SAFETY.PII_LEAK are both severity 5. These errors trigger immediate alerts and should be treated as blocking issues that prevent the response from being delivered to the user.
Example Diagnosis
{
"error_code": "OUTPUT.FORMAT.SCHEMA_VIOLATION",
"confidence": 0.88,
"root_cause": "Agent returned a string where the downstream consumer expected a JSON object with 'result' and 'status' fields.",
"suggested_fixes": ["prompt", "guard"]
}