vault
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill includes a TypeScript code example that reads the Kubernetes ServiceAccount token from its standard sensitive path (
/var/run/secrets/kubernetes.io/serviceaccount/token). While this is a common practice for Kubernetes authentication, it involves direct access to sensitive system credentials.- [INDIRECT_PROMPT_INJECTION]: The skill processes data (secrets) retrieved from external Vault secret engines, which represents a vulnerability surface for indirect prompt injection. - Ingestion points: Untrusted data enters the context when reading secrets via the
vaultCLI (e.g.,vault kv get) or thenode-vaultlibrary. - Boundary markers: No delimiters or warnings are provided to prevent the agent from interpreting instructions that might be embedded within secret values.
- Capability inventory: The skill utilizes network access for API communication, file system reads (ServiceAccount tokens), and command-line tool execution (
vault,jq). - Sanitization: The provided examples do not demonstrate escaping or validation of the content retrieved from Vault.
Audit Metadata