deploy-to-hosting
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to prepare project archives and verify deployment status.
- Evidence: Instructions in Section 2 use
zipandgit archiveto package source code, while Section 7 usescurlandgrepto verify that the live website renders expected content. - [INDIRECT_PROMPT_INJECTION]: The skill has a surface for indirect prompt injection because it reads and interprets project configuration files created by users or third parties.
- Ingestion points: The agent reads
package.json,.env, and general source code files to determine build settings and environment requirements (Sections 2, 3, and 5). - Boundary markers: There are no specific boundary markers mentioned to prevent the agent from following instructions that might be embedded in the code or metadata files.
- Capability inventory: The agent can create archives, perform network checks via
curl, and interact with Hostinger's API to manage servers, databases, and environment variables. - Sanitization: While the skill explicitly mandates that secrets must not be repeated or committed, it does not detail sanitization techniques for other project content against prompt injection.
Audit Metadata