api-manager-effects-and-populators
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The populator factories (columnCountPopulator, sumPopulator, countPopulator) accept table and column identifiers as raw strings. If these are dynamically populated from untrusted user input without validation, it creates a significant SQL injection vulnerability surface.\n
- Ingestion points: The 'opts' parameter object in the populator functions within SKILL.md.\n
- Boundary markers: Absent; no logic is present to delimit or validate the identifiers.\n
- Capability inventory: The pattern provides database read and aggregation capabilities.\n
- Sanitization: Absent; the code relies on the developer to provide static or pre-validated strings.\n- [DATA_EXFILTRATION]: The cdnPurge function utilizes a dynamic key for environment variable access: process.env[zoneEnvVar]. Using variable-derived keys to access the environment is a data exposure risk if the key name is influenced by external input, potentially allowing an actor to probe or identify existing environment variable names via the CDN purging interface.
Audit Metadata