NYC

dag-dynamic-replanner

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • Dynamic Execution (MEDIUM): The skill dynamically modifies task configurations and rewires dependency graphs based on runtime triggers. Because it lacks a validation layer for the configurations it generates or the skill IDs it invokes, it could be coerced into executing unauthorized tasks if the feedback loop is compromised.
  • Indirect Prompt Injection (LOW): The skill's logic for adaptive replanning relies on processing unvalidated text data from external triggers. Ingestion points: The ReplanTrigger object, specifically the 'reason' and 'suggestedAction' fields. Boundary markers: No delimiters or explicit instructions are present to isolate untrusted external data from the modification logic. Capability inventory: The skill possesses the Write, Edit, and Task tools, allowing it to modify and execute workflow configurations. Sanitization: There is no evidence of sanitization for input data before it is incorporated into new task nodes or configurations.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 06:05 PM