neo4j-snowflake-graph-analytics-skill
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCREDENTIALS_UNSAFEPRIVILEGE_ESCALATIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read and process data from Snowflake tables and views to prepare graph projections, creating an attack surface where malicious data could influence the agent. Ingestion points include multiple SQL queries in SKILL.md. Boundary markers are absent, and no sanitization of the retrieved data is specified. The agent possesses capabilities including Bash, WebFetch, and Snowflake SQL execution.
- [CREDENTIALS_UNSAFE]: The skill describes a workflow for managing Snowflake Programmatic Access Tokens (PATs) where the agent is instructed in SKILL.md to 'collect the token from the user' and use it in a SQL SECRET object, involving the agent in a sensitive credential-handling process.
- [PRIVILEGE_ESCALATION]: The instructions for application setup utilize the high-privilege ACCOUNTADMIN role for tasks such as creating roles and granting application permissions, as seen in SQL blocks within SKILL.md.
- [EXTERNAL_DOWNLOADS]: The README contains installation instructions that fetch the skill from a GitHub repository using npx. This involves executing remote content from a vendor-controlled source.
Audit Metadata