levelup-specify
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests the current agent session to extract reusable patterns, creating a potential surface for indirect prompt injection. * Ingestion points: The current agent session (Phase 1). * Boundary markers: No explicit boundary markers or 'ignore' instructions are used when reviewing session content. * Capability inventory: The skill performs git operations (diff, log, status) and writes markdown files to the local filesystem. * Sanitization: There is no explicit sanitization step for session data before it is recorded in CDR drafts.
- [COMMAND_EXECUTION]: Local shell and PowerShell scripts are used to resolve project paths and configuration. * Evidence: Executes scripts/bash/setup-levelup-specify.sh and scripts/powershell/setup-levelup-specify.ps1. * Evidence: Uses git commands to collect implementation evidence.
- [DYNAMIC_EXECUTION]: The setup script employs an inline Python block to format its final status as JSON. * Evidence: python3 call within the bash setup script.
Audit Metadata