writing-kit-skills
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides the agent with specific shell commands to execute locally for maintenance and validation tasks. This includes a specialized awk command for word count verification and several bash and python3 scripts located in the tools/ directory for catalog building and scoring evaluations.
- [INDIRECT_PROMPT_INJECTION]: The skill acts on other instruction files (SKILL.md) which are potential vectors for injection. However, the skill explicitly instructs the agent to treat repository text as 'evidence' rather than 'instructions' and provides a robust framework for auditing and redaction, effectively mitigating this risk.
- [SAFE]: The skill demonstrates high security awareness by mandating the redaction of credentials (tokens, keys, passwords) and forbidding the use of cloud-based agents for sub-tasks, favoring local execution instead.
Audit Metadata