a-plus-onboard-correction

Warn

Audited by Gen Agent Trust Hub on Mar 9, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Generates administrative Ruby scripts intended for execution on target servers. These scripts use ActiveRecord for direct database manipulation, including create!, update_all, and soft-delete (discard) operations.
  • [COMMAND_EXECUTION]: Scripts interact with the application's background processing infrastructure by enqueuing Sidekiq jobs (AttendancePlus::Job::SyncSchoolAttendance.perform_async).
  • [REMOTE_CODE_EXECUTION]: The skill's primary function is dynamic code generation. It constructs complete Ruby scripts that load the full application environment using require_relative '../config/environment'.
  • [COMMAND_EXECUTION]: The skill is explicitly designed to bypass application-level safety guards (such as the tiers_mutable? check mentioned in references/tiers.md) to perform restricted database updates via direct ActiveRecord calls.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection where malicious user input could influence the generated script logic.
  • Ingestion points: User requests provided as "Business Language" (described in SKILL.md).
  • Boundary markers: None; the skill lacks delimiters or instructions to ignore embedded commands in user input.
  • Capability inventory: Perform database writes (create!, update_all), soft-deletes, and Sidekiq job enqueuing across all generated scripts.
  • Sanitization: None; user-provided strings and values are interpolated directly into Ruby code templates.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 9, 2026, 12:38 AM