solutions-installer

Pass

Audited by Gen Agent Trust Hub on Aug 6, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill clones the solution repository from https://github.com/Snowflake-Labs/sf-solutions.git to a local temporary directory. This repository contains the solution manifests, READMEs, and SQL scripts required for installation.
  • [COMMAND_EXECUTION]: Uses the Bash tool to execute git clone for retrieving the remote repository and git pull for updates. This is a standard operation for fetching deployment materials from the official vendor.
  • [REMOTE_CODE_EXECUTION]: The skill reads SQL setup and teardown scripts (install_scripts and teardown_scripts) from the cloned repository and executes them using the snowflake_sql_execute tool. This behavior is the primary intended function of the skill and targets an official vendor source.
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests and executes SQL scripts from an external repository.
  • Ingestion points: SQL scripts and manifest files are fetched from the Snowflake-Labs/sf-solutions GitHub repository.
  • Boundary markers: The skill reads specific scripts defined in a structured manifest.json file and presents an installation plan for user confirmation before execution.
  • Capability inventory: The skill uses snowflake_sql_execute to run database commands and Bash for repository management.
  • Sanitization: The skill executes the SQL content provided in the repository files directly. Because the source is an official vendor repository, the risk is consistent with standard solution deployment.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 6, 2026, 03:06 AM
Security Audit — agent-trust-hub — solutions-installer