db-load-dt

Pass

Audited by Gen Agent Trust Hub on Sep 7, 2026

Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes 1C:Enterprise platform binaries (1cv8.exe, ibcmd.exe) to perform database restoration. This is the primary function of the tool and is performed using standard subprocess calls. Evidence: Found in scripts/db-load-dt.ps1 via Invoke-PlatformProcess and scripts/db-load-dt.py via run_v8 and run_ibcmd functions.
  • [DYNAMIC_EXECUTION]: The scripts dynamically resolve platform paths and construct CLI arguments based on user-provided parameters. Evidence: Both PowerShell and Python scripts include validation logic (Assert-ExtraArgs) that prevents users from supplying specific batch commands that could override or hijack the skill's intended operation.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes path references to external data (DT files) and accepts CLI arguments that influence subprocess execution, creating a potential injection surface. 1. Ingestion points: Path to the input .dt file and user-supplied additional arguments in SKILL.md. 2. Boundary markers: The skill instructions explicitly mandate the agent to seek user confirmation and provide clear warnings about the destructive nature of the database overwrite. 3. Capability inventory: Subprocess execution of the 1C platform for database management. 4. Sanitization: The scripts implement a redaction mechanism for credentials and a whitelist-style filter for CLI arguments to ensure only intended parameters are passed to the underlying platform.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 7, 2026, 10:29 PM
Security Audit — agent-trust-hub — db-load-dt