analyzing-schema-change-storage-risk
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides a shell command using
curlto retrieve storage metrics from a CockroachDB status endpoint. The command uses the-k(insecure) flag, which bypasses SSL certificate verification; while common in local database administration, it is a known security trade-off. - [INDIRECT_PROMPT_INJECTION]: The instructions incorporate placeholders (
<table>and<node>) that are interpolated into shell and SQL commands. This creates an ingestion surface for untrusted data. - Ingestion points: User-provided table names and node addresses in SKILL.md.
- Boundary markers: None present.
- Capability inventory: Shell execution (
curl) and SQL execution (SHOW RANGES,SHOW JOBS). - Sanitization: None present; the skill assumes the agent or user provides valid inputs.
Audit Metadata