solutions-installer

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.85). The skill clones/reads the outsider-authored sf-solutions repository at runtime (public GitHub content) and ingests its free-text manifest.json/README.md plus SQL script text into the agent’s execution context (Step 1–3), which can include indirect prompt-injection payloads.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The skill clones and reads runtime content from https://github.com/Snowflake-Labs/sf-solutions.git (git clone ...) and then executes SQL scripts and behavior defined in that repo (manifest.json and scripts) via snowflake_sql_execute, so fetched remote content directly controls executed code.

MEDIUM W021: Hidden or invisible Unicode characters detected (potential obfuscation or prompt injection).

• Hidden Unicode characters detected (1 type(s) found)