convex
Pass
Audited by Gen Agent Trust Hub on Mar 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the user to add functions to
scripts/setup.tsthat execute shell commands viaspawn, includingconvex dev,bun convex env, andlsof. These commands are used to manage the local development backend. - [PROMPT_INJECTION]: The skill contains an indirect prompt injection surface in the provided setup scripts.
- Ingestion points: Environment variables such as
SITE_URLare read fromprocess.envwithin the generatedscripts/setup.ts(defined inflows/setup/FLOW.md). - Boundary markers: Absent; values from the environment are passed directly as arguments to the
spawnfunction without delimiters or isolation. - Capability inventory: The script in
scripts/setup.ts(defined inflows/setup/FLOW.md) has the capability to execute subprocesses viaspawnand kill processes viaprocess.kill. - Sanitization: No validation or sanitization is performed on the environment variable values before they are used in command execution.
Audit Metadata