linkerd-expert
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches the Linkerd CLI installation script from the official Linkerd project domain (run.linkerd.io).
- [REMOTE_CODE_EXECUTION]: Includes instructions for piping a remote shell script directly into the shell for CLI installation (curl | sh).
- [COMMAND_EXECUTION]: Extensive use of shell commands through kubectl and linkerd tools for cluster management and mesh configuration, consistent with the allowed-tools configuration.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect instructions through data ingestion. 1. Ingestion points: Processes live network traffic data from 'linkerd viz tap', proxy logs, and OpenAPI specifications. 2. Boundary markers: None identified in the prompt instructions to delineate or ignore potential instructions in untrusted data. 3. Capability inventory: The skill has the ability to execute cluster-modifying commands such as 'kubectl apply' and 'linkerd install'. 4. Sanitization: No explicit instructions for validating or sanitizing the content of tapped traffic or imported resource definitions.
Audit Metadata