setup
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- Command Execution: The skill utilizes the Vercel CLI to perform administrative tasks such as enabling CI, listing runs, and modifying repository settings.
- Evidence: Use of
vercel api,vercel ci enable, andvercel ci logscommands inreferences/vercel-ci.mdto interact with Vercel's infrastructure. - Indirect Prompt Injection Surface: The skill is instructed to read and analyze build logs from failed CI tasks to provide explanations and fixes. This represents a potential surface where external data (build output) is brought into the agent's context.
- Ingestion points: Build logs retrieved via the
vercel ci logscommand inreferences/vercel-ci.mdandSKILL.md(Step 6). - Boundary markers: No specific delimiters or "ignore instructions" wrappers are defined for the log content.
- Capability inventory: The agent has the ability to execute Vercel CLI commands, perform Git operations (branching, publishing), and open GitHub pull requests.
- Sanitization: The instructions do not specify sanitization or filtering for the external log content before it is processed by the agent.
Audit Metadata