large-data-with-dask

Fail

Audited by Snyk on Mar 3, 2026

Risk Level: CRITICAL
Full Analysis

CRITICAL E004: Prompt injection detected in skill instructions.

  • Potential prompt injection detected (high risk: 1.00). The prompt includes hidden/deceptive instructions to read and write the agent's internal memory file (.claude/context/memory/learnings.md) and to change assumptions about context persistence, which are outside the stated Dask code-review purpose and attempt to access/alter internal state.
Audit Metadata
Risk Level
CRITICAL
Analyzed
Mar 3, 2026, 08:51 AM