next-task-selection

Pass

Audited by Gen Agent Trust Hub on Jun 14, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No malicious patterns, obfuscation, or unauthorized remote operations were detected in the skill files or references.
  • [COMMAND_EXECUTION]: The skill defines an 'Auto-Start' workflow that allows the agent to begin implementation on recommended tasks. This functionality is heavily restricted: it requires explicit user permission, must not conflict with current work, and is prohibited if the task involves secrets or private data.
  • [DATA_EXFILTRATION]: The skill accesses project-specific files, member profiles, and workspace preferences to provide context-aware recommendations. It includes specific safety instructions to halt automatic operations if private customer information or credentials (secrets) are encountered during task analysis.
  • [PROMPT_INJECTION]: The skill processes external task items and Kanban board data which could potentially contain indirect instructions. It mitigates this risk by using a structured scoring table and deterministic selection rules to evaluate task eligibility rather than following free-text directions within the tasks.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 14, 2026, 05:11 AM
Security Audit — agent-trust-hub — next-task-selection