find-fallacies

Pass

Audited by Gen Agent Trust Hub on Feb 19, 2026

Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
  • [Indirect Prompt Injection] (LOW): The skill processes untrusted data, which is a common surface for indirect prompt injection. * Ingestion points: Analyzes 'the provided text' as described in SKILL.md. * Boundary markers: Absent; there are no instructions to use delimiters or ignore embedded commands in the input. * Capability inventory: None; the skill does not use any subprocesses, file operations, or network calls. * Sanitization: Absent.
  • [No Code] (SAFE): This skill contains only markdown instructions and metadata. No scripts, binaries, or configuration files for code execution are present.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 19, 2026, 04:22 PM