skills/canner/wren-engine/wren-sql/Gen Agent Trust Hub

wren-sql

Pass

Audited by Gen Agent Trust Hub on Mar 20, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Periodically checks for updates by retrieving a version manifest from the project's official repository on raw.githubusercontent.com.
  • [REMOTE_CODE_EXECUTION]: Provides instructions to update the skill using a shell script downloaded from the official GitHub repository and piped to bash. This is a standard practice for developer tool maintenance.
  • [COMMAND_EXECUTION]: Includes shell commands for manual maintenance and updating of the SQL generation environment.
  • [PROMPT_INJECTION]: The version check mechanism represents a potential surface for indirect instructions.
  • Ingestion points: Silently fetches JSON data from a remote URL (SKILL.md).
  • Boundary markers: Absent; remote data is processed directly for version string comparison.
  • Capability inventory: Recommends shell command execution (SKILL.md) and handles complex SQL logic across multiple files (SKILL.md, references/bigquery.md, references/correction.md, references/datetime.md, references/types.md).
  • Sanitization: None; the remote metadata is used directly without validation beyond string comparison.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 20, 2026, 08:56 AM