cloud-ship
Pass
Audited by Gen Agent Trust Hub on Nov 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a shell command using
sedto extract the version number from the skill's ownSKILL.mdfile for reporting purposes. This is a benign use of system utilities for self-metadata retrieval. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and relay output from the
shipskill verbatim, which introduces a potential surface for indirect prompt injection if the source skill's output is compromised. 1. Ingestion points: The agent reads the 'step-5 result' and 'merge summary' produced by theshipskill. 2. Boundary markers: No explicit delimiters or instructions to ignore embedded commands in the relayed content are provided. 3. Capability inventory: The skill invokes theshiptool and performs text processing withsed. 4. Sanitization: Output is relayed verbatim without sanitization or validation steps.
Audit Metadata