task-management
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The
router.shscript executes local TypeScript files (task-cli.ts) usingnpx ts-nodeto provide a functional command-line interface. This execution is limited to internal skill scripts and does not involve remote or untrusted code sources. - [INDIRECT_PROMPT_INJECTION]: The skill reads task definitions and summaries from JSON files in the
.tmp/tasks/directory (ingestion points:scripts/task-cli.ts). There are no specific boundary markers for the data when outputted to the agent. The skill has the capability to write to the local filesystem within the task directory (capability inventory:scripts/task-cli.ts). It performs sanitization by enforcing a 200-character limit on task summaries.
Audit Metadata