chimera_detect
Detects hallucinations and security threats in AI outputs using configurable strategies like range, dictionary, and cross-reference validation. Returns hallucination flags and injection signals.
Instructions
Hallucination and MCP security detection. Strategies: range, dictionary, semantic, cross_reference, temporal, confidence_threshold. Returns hallucination flags plus prompt-injection/tool-poisoning signals.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| value | Yes | Value to scan | |
| confidence | No | ||
| strategy | Yes | ||
| params | No | Strategy params: range: {valid_range: [min, max]} dictionary: {allowed_values: [...]} semantic: {forbidden_patterns: [...]} (omit for default absolute-certainty scan) cross_reference: {reference_values: [...], tolerance: 0.1} temporal: {max_age_seconds: 3600, reference_time: <unix_ts>} confidence_threshold: {threshold: 0.7} |