SecEng Defend · Labs
Output Handling Safety Tester
Paste model output, select the output sink, and get instant safety analysis. Tests HTML rendering, Markdown link safety, JSON parsing risks, tool call arguments, and side-effect patterns.
- • HTML sink: script tags, event handlers, iframe, object/embed detection
- • Markdown sink: image beacons, javascript: links, open redirects
- • JSON sink: command fields, over-scoped resource arguments
- • Tool call sink: dangerous action patterns, argument injection indicators
- • All sinks: hidden context leakage, embedded secret detection
8 sink types
HTML, Markdown, JSON, tool call, email, DB, code
15 output rules
Across hidden leakage, injection, side effects
OWASP LLM05
Improper output handling mapped
No LLM calls
Fully deterministic pattern detection
Load example:
Next step
Need a full output handling review?
We assess model output sinks, rendering pipelines, injection defenses, and side-effect authorization controls — mapped to OWASP LLM Top 10 and compliance frameworks.