delegate-imagegen
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill acts as a bridge between a main agent and external sub-processes, creating a surface where malicious instructions in user-provided content could influence the behavior of the sub-agent.
- Ingestion points: User-provided Markdown content containing task objectives, scope, and context is accepted via stdin and passed to the delegation logic through
run-imagegen.shandprepare-imagegen.sh. - Capability inventory: The skill's shims execute Node.js commands and standard shell utilities (
ls,file,test,bash) to manage files, directories, and process execution. - Boundary markers: The skill instructions specify a structured Markdown format (Objective, Scope, Context, etc.) to organize the request data.
- Sanitization: No explicit validation or escaping of the input Markdown is performed within the shell wrappers before being passed to the execution environment.
- [COMMAND_EXECUTION]: The skill utilizes multiple shell scripts that execute a local Node.js script (
delegate-cli.mjs) to perform task delegation logic. This execution is local to the skill's own directory and is used to manage the lifecycle of sub-processes and file system operations related to the delegation tasks.
Audit Metadata