add-table-to-offline-profile
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFE
Full Analysis
- Command Execution: The skill utilizes Bash to run localized utility scripts (e.g.,
dataverse-request.js) and Node.js for parsing configuration files. It explicitly mandates shell-quoting for all invocation values to mitigate command injection risks. - Environment and Scope Validation: Before performing operations, the skill verifies that the
environmentIdin the localpower.config.jsonmatches the orchestrator-provided context. It also enforces that target tables must exist within an approved allowlist. - Path Enforcement: The skill uses absolute paths for both project files (relative to
<working_dir>) and internal scripts (relative to${PLUGIN_ROOT}), preventing path traversal or accidental execution from the current working directory. - Indirect Prompt Injection Surface: The skill ingests data from project configuration files and performs greps on source code. While these are external inputs, the risk is managed through input validation and the fact that model invocation is disabled within this skill's configuration.
Audit Metadata